Multi-node space-time synchronous scheduling method supporting distributed co-simulation

Abstract

The invention provides a multi-node space-time synchronous scheduling method supporting distributed co-simulation, and relates to the technical field of ship scheduling, and the method comprises the steps: obtaining a historical distribution probability of each factor type according to a ship type, a navigation influence coefficient and a starting destination in historical data; the simulation occurrence probability of each factor type is calculated by fusing random numbers and serves as input data of a ship running simulation model, a dispatching model is constructed through a graph neural network, ships are dispatched, simulation ships are established, and according to the ship running simulation model, the navigation speed and position of the simulation ships are output and serve as input of the dispatching model; and scheduling the simulation ship, calculating a comprehensive evaluation value of the scheduling area according to the navigation speed, the navigation time, the navigation distance and the ship position of the simulation ship scheduled by the scheduling model, and scheduling the ship in real time after the comprehensive evaluation is satisfied.

Description

Technical Field

[0001] This invention relates to the field of ship scheduling technology, specifically to a multi-node spatiotemporal synchronization scheduling method that supports distributed collaborative simulation. Background Technology

[0002] In ship scheduling systems over the past few decades, ship navigation scheduling has typically relied on single static models or empirical rules. These methods have significant limitations when dealing with dynamic and complex maritime environments. Especially when multiple ships operate simultaneously and the maritime environment is complex and unpredictable, traditional scheduling methods struggle to effectively consider the interactions between ships, environmental changes, and the real-time impact of scheduling commands on ship navigation. Furthermore, traditional systems often neglect the importance of spatiotemporal synchronization and multi-node collaborative simulation, failing to provide real-time and accurate navigation status predictions and scheduling optimizations for decision-making.

[0003] To improve the intelligence and accuracy of ship scheduling systems, research on distributed simulation and collaborative scheduling has gradually emerged in recent years. Combining spatiotemporal synchronization and distributed simulation technologies enables ship scheduling on a larger scale and with higher precision. However, achieving efficient spatiotemporal synchronized scheduling, accurate simulation deduction, and optimized scheduling decisions in multi-node distributed systems remains a challenge for current technology.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-node spatiotemporal synchronization scheduling method that supports distributed collaborative simulation, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation includes the following steps: Step 1: Obtain the historical navigation data, environmental data, and route data of each ship in the area to be dispatched, and stamp them with timestamps. Calculate the historical navigation impact coefficient of each ship at the same moment. Step 2: Construct a ship navigation simulation model. Use different ship types, navigation impact coefficients, route data, origin and destination, and dispatch instructions from historical data as inputs to the ship navigation simulation model, and the ship's speed and position as outputs to train the ship navigation simulation model. Step 3: Obtain the historical distribution probability of each factor type based on the ship type, navigation influence coefficient, and origin and destination from the historical data. Calculate the simulation occurrence probability of each factor type by fusing the historical distribution probability with random numbers. Generate simulation data for each factor type based on the simulation occurrence probability and input it into the trained ship navigation simulation model. Output the simulated ship's navigation speed and position. Step 4: Construct a scheduling input dataset with ships as nodes and the distance between ships as edges. Based on the impact of scheduling instructions on ship navigation, construct a scheduling model using a graph neural network to schedule ships. Step 5: Based on the number of ships in the simulation data, create the same number of simulated ships. Input the scheduling instructions that are sent at the same time as the simulation data into the ship navigation simulation model for simulation. The ship navigation simulation model outputs the sailing speed and position of the simulated ships as the input of the scheduling model to schedule the simulated ships. Step 6: Based on the sailing speed, sailing time, sailing distance, and ship position of the simulated ships after scheduling by the scheduling model, calculate the comprehensive evaluation value of the scheduling area by judging the congestion coefficient of each simulated ship, and optimize the scheduling model parameters of the scheduling model according to the comprehensive evaluation value to perform real-time scheduling of ships after comprehensive evaluation.

[0007] Furthermore, the environmental data includes wind speed, wind direction, water flow velocity, and water flow direction; The route data includes ship traffic density and sailing intervals; The navigation data includes ship parameters, speed, direction of travel, and position. Ship parameters include tonnage and power; The specific steps for calculating the historical navigation influence coefficient of each ship at the same moment are as follows: The mass-energy product of a ship is determined based on its tonnage and power. The ship's parameter factors are determined by weighting the mass-energy product and its speed. By using environmental data, specifically the angle between wind speed, wind direction and the ship's direction of travel, we can determine the disturbance of wind on the ship's travel; by using the angle between water flow speed, water flow direction and the ship's direction of travel, we can determine the disturbance of water flow on the ship's travel; and by assigning weights to the disturbances of wind and water flow, we can calculate the environmental disturbance factor. By using route data, specifically the relationship between the angle and distance between other vessels and the route, combined with the impact of spatial attenuation on vessels, and considering the impact of route density on navigation, the route congestion factor is calculated. The navigation impact coefficient is calculated by comprehensively considering ship parameter factors, environmental disturbance factors, and route congestion factors, and by introducing nonlinear coupling effects. The specific calculation formula is as follows: Among them, among them, For ship parameter factors, As environmental disturbance factors, For route congestion factor, This is calculated as a navigation impact factor. Coupling index, This is a correction item for the ship's anti-interference capability.

[0008] Furthermore, the ship navigation simulation model includes an input state vector, a strategy function, a ship state update function, and a motion vector; The ship navigation simulation model includes an input state vector, a strategy function, a ship state update function, and a motion vector.

[0009] Specifically, the ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions are used to construct the state vector of the simulation model as the input of the model, while the navigation speed and direction are used to construct the action vector as the output of the model. The ship state update function includes a position update function and a velocity update function; The system integrates multi-dimensional data such as ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions into a comprehensive state vector. This state vector serves as the basis for the ship navigation simulation model. The real-time input provides a comprehensive picture of the ship's current environment and mission status; The policy function makes decisions based on the state vector. It calculates the probability distribution of all possible action vectors in the current state through the embedded probability distribution function, and samples or selects the optimal action accordingly. After the motion vector is output, the ship state update function begins to work, updating the ship state according to the laws of physical motion: The position update function uses the current velocity, direction, and navigation influence coefficient to calculate the position at the next moment, while the velocity update function calculates the new velocity based on the acceleration command output by the strategy function, the current direction, and the velocity decay factor, thus completing the advancement of one simulation step.

[0010] Furthermore, the calculation steps for calculating the simulation occurrence probability of each factor type by fusing historical distribution probabilities with random numbers are as follows: The method for obtaining the historical distribution probability of ship type and quantity is as follows: Within the dispatch area, by time interval The number of ships appearing within the historical time frame is statistically analyzed by type, and the probability of each type of ship appearing is calculated. The method for obtaining the historical distribution probability of the navigation impact coefficient is as follows: Based on the same time interval Calculate the navigation impact coefficient for ships within the area to be dispatched, based on the historical time length. * Within this, the navigation impact coefficient for each ship is calculated, along with the probability of each navigation impact coefficient value occurring. The method for obtaining the historical distribution probability of the starting and destination locations is as follows: Within the area to be dispatched, by time interval The statistics of the starting positions of occurrences within the historical time frame are compiled, and the probability of occurrence at each starting position is calculated.

[0011] Furthermore, the specific calculation formula for calculating the simulation occurrence probability of each factor type by fusing historical distribution probabilities with random numbers is as follows: in, For the first The probability of occurrence of each factor type in the simulation. For the first Historical distribution probability of each factor type The maximum random number, The random fluctuation factor follows a uniform distribution with a distribution range of [value missing]. .

[0012] Furthermore, the scheduling input dataset also includes node features and edge features. Node characteristics include driving data and scheduling instructions; The edge feature is the distance between ships; The construction of a scheduling model includes scheduling model establishment and scheduling model training; The scheduling model establishment includes spatial aggregation function, temporal evolution function, spatial attention function, and instruction output function; The process of training the scheduling model involves using historical scheduling instructions and driving data as inputs to the model, and using sailing speed and position as outputs to schedule the model.

[0013] Further, the specific steps of step 4 are as follows: update the simulation data according to the same time interval, establish the same number of simulated ships according to the number of ships in the generated simulation data, schedule all simulated ships at the same time, input the updated simulation data and scheduling instructions into the ship driving simulation model, output the sailing speed and position of the simulated ships, use the sailing speed and position of the simulated ships as input to the scheduling model, schedule all simulated ships, and issue scheduling instructions to the ship driving simulation model to complete one simulation scheduling process.

[0014] Furthermore, the congestion coefficient is calculated as follows: Based on the simulated ships' speed and the distance between them, the congestion coefficient for each simulated ship is calculated using the following formula: in, For the first Congestion coefficient of each ship, For ships With ships The distance between them Ships With ships sailing speed, For ships Number of ships in the vicinity.

[0015] Furthermore, the calculation steps for the comprehensive evaluation value are as follows: By dividing the route into equal intervals, segments are formed based on these equal intervals. The number of ships in each segment is counted, with the segment as the horizontal axis and the number of ships as the vertical axis. The segment where the peak number of ships in each segment is located in the positive direction of the horizontal axis is identified as a ship congestion area. A comprehensive evaluation value is calculated by calculating the number of ships and the congestion coefficient between each area. The method for calculating the comprehensive evaluation value is as follows: in, For comprehensive evaluation, For the first in the congested area Congestion coefficient of each ship, The number of ships in the congested area. For the first A congested area The number of congested areas. This represents the average actual arrival time. The planned arrival time, This represents the average distance actually traveled. This represents the average distance the plan is to travel.

[0016] Furthermore, the method for optimizing the scheduling model parameters based on the comprehensive evaluation value is to perform optimization through a sparrow search algorithm, specifically including a discoverer and a follower.

[0017] Compared with the prior art, the beneficial effect of the present invention is that the present invention obtains the historical distribution probability of each factor type based on the ship type, navigation influence coefficient and origin and destination from historical data, and integrates random numbers to calculate the simulation occurrence probability of each factor type as input data of the ship driving simulation model. The scheduling model is constructed through graph neural network to schedule ships and establish simulated ships. According to the ship driving simulation model, the sailing speed and position of the simulated ships are output as input to the scheduling model to schedule the simulated ships. According to the sailing speed, sailing time, sailing distance and ship position of the simulated ships after scheduling by the scheduling model, the comprehensive evaluation value of the scheduling area is calculated, and the ships are scheduled in real time after the comprehensive evaluation is satisfied. This invention calculates the ship navigation impact coefficient based on historical voyage data, environmental data, and route data, and performs real-time simulation of ship navigation through a simulation model. This enables ship scheduling to be dynamically adjusted according to the real-time environment and ship status, thereby avoiding the inefficiency and lag problems of traditional methods. Furthermore, by employing a graph neural network scheduling model that combines the distance between ships, scheduling instructions, and navigation influencing factors, intelligent ship scheduling decisions were achieved. This model, by optimizing the simulated ship speed, position, and congestion coefficient, not only improves scheduling efficiency but also optimizes navigation routes and times, reducing ship conflicts and congestion during navigation. In addition, based on real-time feedback from simulation data and scheduling instructions, the scheduling model parameters can be dynamically optimized, ensuring spatiotemporal consistency and high efficiency of collaborative scheduling during the scheduling process. This improves the overall effectiveness and intelligence level of ship scheduling and effectively solves the problems of poor spatiotemporal synchronization and dynamic environment adaptability in traditional ship scheduling systems. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: Please see Figure 1 The present invention provides a technical solution: A multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation includes the following steps: Step 1: Obtain the historical navigation data, environmental data, and route data of each ship in the area to be dispatched, and stamp them with timestamps. Calculate the historical navigation impact coefficient of each ship at the same moment. Step 2: Construct a ship navigation simulation model. Use different ship types, navigation influence coefficients, route data, origin and destination, and dispatch instructions from historical data as inputs to the ship navigation simulation model, and ship speed and position as outputs to train the ship navigation simulation model.

[0022] The navigation impact coefficient is a key indicator that integrates ship parameters, environmental factors, and route conditions. It reflects the degree of influence of various factors on the ship's navigation state and directly affects the simulation model of ship navigation. By integrating the ship's basic parameters, environmental data, and route data, it helps simulate the ship's real behavior in complex environments, making the simulation results closer to reality. This is crucial for multi-node spatiotemporal synchronous scheduling, ensuring that each node (ship) maintains consistent dynamic changes and interactions in collaborative simulation. By considering the impact of various factors on ship navigation, the navigation impact coefficient can provide accurate input data for the scheduling model. For example, in densely populated areas, route congestion factors may significantly affect ship navigation efficiency. The scheduling system can adjust routes or ship speeds and paths accordingly to avoid congestion and improve efficiency. Spatiotemporal synchronization between ships is the foundation of distributed collaborative scheduling. By calculating accurate navigation impact coefficients, it can be ensured that the ship's navigation process reflects changes in the environment and adjacent ships in real time during multi-node simulations. In this way, the scheduling of different ships can be carried out within the correct spatiotemporal synchronization framework, avoiding conflicts and errors.

[0023] In this embodiment, the environmental data includes wind speed, wind direction, water flow speed, and water flow direction; Route data includes vessel traffic density and sailing intervals; The navigation data includes ship parameters, speed, direction of travel, and position. Ship parameters include tonnage and power; The specific steps for calculating the historical navigation influence coefficient of each ship at the same moment are as follows: The mass-energy product of a ship is determined based on its tonnage and power. The ship's parameter factors are determined by weighting the mass-energy product and its speed. The specific calculation formula is as follows: in, For ship parameter factors, For ship tonnage, For ship power, For ship speed, These represent the mass-energy product and the weighting of sailing speed, respectively. .

[0024] Ship tonnage directly represents the ship's mass. The greater the mass, the greater the inertia, and the more difficult it is to accelerate or decelerate. Ship power represents the ship's ability to overcome inertia. The greater the power, the greater the ship's "potential" to change its speed. The use of an exponential function here is a key design feature because it non-linearly amplifies the value of high power. This means that for two ships of the same tonnage, the one with greater power will have an exponential increase in its ability to overcome inertia and achieve maneuverability, rather than just a linear increase. This is more in line with engineering practice because a significant increase in power will significantly improve the ship's maneuverability.

[0025] Therefore, by constructing the concept of "mass-energy product": the product of mass and the energy potential to overcome inertia, this value comprehensively reflects the potential influence of the ship's inherent properties on its dynamic behavior.

[0026] Ship parameter factors consider a ship's basic performance, tonnage, and power. A ship's tonnage and power determine its dynamic characteristics during navigation, directly impacting its speed and range. By calculating the ship's mass-energy product based on its tonnage and power, and adjusting the ship parameter factors using a weighted approach, the performance differences between different ship types can be reflected. This allows for more accurate simulation of ship behavior under various environmental conditions, ensuring that the scheduling model can accurately predict the behavior of each ship in a multi-ship system.

[0027] By using environmental data—specifically, the angle between wind speed, wind direction, and the ship's direction of travel—the disturbance of wind on the ship's movement is determined. Similarly, the disturbance of water flow on the ship's movement is determined by the angle between water flow speed, water flow direction, and the ship's direction of travel. The environmental disturbance factor is calculated by assigning weights to the disturbances of wind and water flow. The specific calculation formula is as follows: in, As environmental disturbance factors, For wind speed, For water flow velocity, For wind direction, In terms of water flow direction, For the direction of the ship's travel, The wind speed influence coefficient, The coefficient representing the influence of water flow velocity. + .

[0028] Environmental disturbance factors mainly originate from the external environment, such as wind speed, wind direction, water flow velocity, and water flow direction. These factors directly affect the navigation status of ships, especially when wind speeds are high or water flow velocities are strong, which can significantly disrupt the ship's speed and path. By calculating environmental disturbance factors, the impact of the environment on ship navigation can be accurately reflected, enabling simulation models to simulate the ship's navigation status under adverse weather or complex water flow conditions, thereby improving the accuracy and reliability of the simulation.

[0029] Specifically, for wind disturbance terms Where FV is the wind speed scalar. The cosine of the angle between the wind direction and the ship's heading was calculated. Physically, this represents the projection of the wind speed vector onto the ship's heading axis. A positive result indicates tailwind thrust, a negative result indicates headwind drag, and a result of zero indicates no axial velocity influence from crosswinds. Similarly, the water flow disturbance term... The projection effect of the current velocity vector onto the heading direction is evaluated using the same mechanism. Finally, through weighting... and The two projected values ​​are weighted and merged, with the weights reflecting the relative importance of wind and water flow on the impact of ships in a specific aquatic environment (for example, the influence of wind may be higher in open seas, while the influence of water flow may be dominant in narrow waterways), thus obtaining a comprehensive and directional environmental disturbance factor. A positive value indicates that the overall environment helps the ship increase speed, while a negative value indicates that the overall environment hinders the ship from decelerating. This factor provides a key environmental dynamic input for the subsequent simulation model, ensuring that the ship's response to external disturbances in the virtual environment conforms to physical laws and has quantitative accuracy.

[0030] By analyzing route data—specifically, the relationship between the angle and distance between other vessels and their positions on the route—and considering the impact of spatial attenuation on vessels, as well as the influence of route density on navigation, the route congestion factor is calculated. The specific calculation formula is as follows: in, For route congestion factor, For route density, The distance between ships The spatial attenuation coefficient, The angle between the ship and the course. .

[0031] The route congestion factor reflects the congestion level of ships on a specific route or channel. This includes factors such as ship traffic density and spacing. If a channel is congested, ships may need to adjust their speed and path to avoid collisions or increase travel time. By calculating the route congestion factor, the scheduling system can assess the current level of congestion on the route in real time and adjust the ship's route or speed according to the actual situation to ensure smooth and safe navigation.

[0032] Specifically, through The calculation of the individual interaction risk between the target vessel and a nearby vessel uses the distance between vessels as the core variable of the exponential decay term, ensuring that the blocking effect decreases sharply with increasing distance, consistent with physical reality (vessels at long distances have almost no impact). This introduces the key factor of relative heading, when the two ships are heading in the same direction ( The risk is greatest when the angle is approximately 0° and cos≈1 (prone to rear-end collisions or parallel congestion), and the heading is perpendicular ( When the angle is 90° and cos=0, the risk drops to zero (no interaction). If the heading is opposite (∆α=180° but constrained to [-90°, 90°], the absolute value effect is taken, and cos is negative), the risk falls between these values. The spatial attenuation coefficient modulates the severity of the entire attenuation process. On the other hand, the route density, as a statistical average, reflects the density of ships in the entire area and is the baseline background for congestion. The function clamps its upper limit to 1, normalizing the factor to represent the relative value of the blocking probability or blocking intensity (0 for no blocking, 1 for complete blocking), thus providing a continuous, smooth, and physically meaningful blocking quantization input for simulation scheduling.

[0033] The navigation impact coefficient is calculated by comprehensively considering ship parameter factors, environmental disturbance factors, and route congestion factors, and by introducing nonlinear coupling effects. The specific calculation formula is as follows: in, For ship parameter factors, As environmental disturbance factors, For route congestion factor, This is calculated as a navigation impact factor. Coupling index, This is a correction item for the ship's anti-interference capability.

[0034] Navigation in the real world is often nonlinear, especially in complex maritime environments. Interactions between ships and between ships and their environment are typically not simple linear relationships. Therefore, employing a nonlinear coupling effect calculation method can more realistically simulate ship performance under different conditions, avoiding errors introduced by simple models and improving the reliability of simulation results. This method, based on distributed collaborative simulation, ensures spatiotemporal synchronization of multiple nodes by calculating accurate navigation influence coefficients. This guarantees that each node (ship) can obtain real-time environmental changes and navigation status, thereby achieving efficient and accurate scheduling optimization. The interactions between ships are accurately simulated, and the system can continuously adjust scheduling strategies in dynamic environments, reducing ship conflicts and resource waste.

[0035] First, the external environmental factors, including environmental disturbance factors and route congestion factors, are treated as a whole, and a nonlinear coupling effect is introduced through the exponential τ. This means that the environmental and congestion factors are not simply superimposed but may aggravate each other (when τ>1) or weaken each other (when τ<1). This simulates the effect of increased risk when severe environmental conditions and traffic congestion coexist in reality. Then, the coupled external influence is multiplied by the parameter factor characterizing the ship's inherent anti-interference capability, reflecting the physical law that "high-horsepower, high-tonnage" ships (high VF value) have stronger resistance when facing the same external challenges.

[0036] In this embodiment, the ship navigation simulation model includes an input state vector, a strategy function, a ship state update function, and an action vector.

[0037] The simulation model's state vector is constructed by considering ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions. As inputs to the model, sailing speed and direction are used to construct the motion vector. ; Specifically, the ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions are used to construct the state vector of the simulation model as the input of the model, while the navigation speed and direction are used to construct the action vector as the output of the model. The ship state update function includes a position update function and a velocity update function; The system integrates multi-dimensional data such as ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions into a comprehensive state vector. This state vector serves as the basis for the ship navigation simulation model. The real-time input provides a comprehensive picture of the ship's current environment and mission status; The policy function makes decisions based on the state vector. It calculates the probability distribution of all possible action vectors in the current state through the embedded probability distribution function, and samples or selects the optimal action accordingly. After the motion vector is output, the ship state update function begins to work, updating the ship state according to the laws of physical motion: The position update function uses the current velocity, direction, and navigation influence coefficient to calculate the position at the next moment, while the velocity update function calculates the new velocity based on the acceleration command output by the strategy function, the current direction, and the velocity decay factor, thus completing the advancement of one simulation step. Policy function: in, Let be the policy function with as the parameter set. For time scale, It is the probability distribution function; Ship status update function: in, This provides the ship's position information at time 𝑡+1. Let $\frac{ ... Let be the direction vector of the ship at time t. For the update function, Let be the navigation impact coefficient at time 𝑡. The output parameter set is the acceleration at time t. This is the velocity decay factor.

[0038] By constructing a ship navigation simulation model, the multi-node spatiotemporal synchronization and scheduling capabilities of distributed collaborative simulation can be significantly improved. The ship navigation simulation model possesses the ability to handle high-dimensional state spaces and complex nonlinear decision-making problems, enabling each ship node to autonomously learn and optimize its navigation strategy based on local environmental perception, thereby achieving intelligent and adaptive dynamic responses in global collaborative scheduling. The input state vector provides real-time perception information, the policy function outputs the optimal action decision, the ship state update function ensures the physical consistency between state evolution and environmental interaction, and the action vector drives the evolution of model behavior. Overall, the ship navigation simulation model not only improves the intelligence level of single-node simulation but also supports efficient spatiotemporal synchronization and behavioral coordination among multiple nodes through reinforcement learning mechanisms, reducing dependence on central control and enhancing the system's distributed autonomy and robustness, thus providing strong technical support for collaborative scheduling in complex maritime traffic environments.

[0039] Step 3: Obtain the historical distribution probability of each factor type based on the ship type, navigation influence coefficient, and origin / destination location from the historical data. Calculate the simulation occurrence probability of each factor type by fusing the historical distribution probability with random numbers. Generate simulation data for each factor type based on the simulation occurrence probability and input it into the trained ship navigation simulation model. Output the simulated ship's navigation speed and position.

[0040] By obtaining the historical distribution probability of each factor type, navigation impact coefficient, and origin / destination location, more accurate and realistic input data can be provided for ship navigation simulation models. Specifically, by calculating the historical distribution probability of ship type and quantity, navigation impact coefficient, and origin / destination location, the occurrence pattern of each factor within a given time range can be obtained, providing a reference for subsequent simulation processes. The historical distribution probability reflects the frequency and probability of different factors occurring in actual scheduling, thus enabling a more realistic simulation of ship operation and reducing the gap between model assumptions and reality. By incorporating random numbers to calculate the simulation occurrence probability of each factor type, the simulation process gains stronger randomness and dynamism, effectively responding to changing external environments. In supporting distributed collaborative simulation and multi-node spatiotemporal synchronous scheduling, this simulation data input method based on historical distribution probability improves the collaborative accuracy and spatiotemporal synchronization among multiple nodes, ensuring that each node can effectively share ship operation information at different time intervals, ensuring the efficiency and accuracy of the overall scheduling method, thereby optimizing the reliability of simulation results and decisions.

[0041] In this embodiment, the calculation steps for calculating the simulation occurrence probability of each factor type by fusing historical distribution probabilities with random numbers are as follows: The method for obtaining the historical distribution probability of ship type and quantity is as follows: Within the dispatch area, by time interval The number of ships appearing within the historical time frame is statistically analyzed by type, and the probability of each type of ship appearing is calculated. The method for obtaining the historical distribution probability of the navigation impact coefficient is as follows: Based on the same time interval Calculate the navigation impact coefficient for ships within the area to be dispatched, based on the historical time length. * Within this, the navigation impact coefficient for each ship is calculated, along with the probability of each navigation impact coefficient value occurring. The method for obtaining the historical distribution probability of the starting and destination locations is as follows: Within the area to be dispatched, by time interval The starting positions of occurrences within the historical time frame are statistically analyzed, and the probability of occurrence at each starting position is calculated. By fusing historical distribution probabilities with random number calculations to determine the simulation occurrence probability of each factor type, a degree of randomness and uncertainty is introduced, making the simulation model more realistic. This method, through statistical analysis of historical data and the introduction of random number bias, simulates the changing trends and uncertainties of factors such as ship type, navigation influence coefficient, and origin / destination location, avoiding overly simplistic or idealized assumptions, resulting in more realistic and dynamic simulation results. In multi-node spatiotemporal synchronization scheduling methods supporting distributed collaborative simulation, this approach ensures that each node can generate different simulation results based on the same historical data, thereby enhancing the overall coordination and reliability of the system. By fusing historical distribution probabilities and introducing random number bias, the spatiotemporal synchronization of multiple nodes is improved, ensuring effective response and information sharing among different nodes during simulation, enhancing scheduling flexibility and adaptability, and contributing to optimizing the overall performance and decision-making effectiveness of the ship navigation simulation system.

[0042] In this embodiment, the specific calculation formula for calculating the simulation occurrence probability of each factor type by fusing historical distribution probabilities with random numbers is as follows: in, For the first The probability of occurrence of each factor type in the simulation. For the first Historical distribution probability of each factor type The maximum random number, The random fluctuation factor follows a uniform distribution with a distribution range of [value missing]. .

[0043] Step 4: Construct a scheduling input dataset with ships as nodes and the distance between ships as edges. Based on the impact of scheduling instructions on ship navigation, construct a scheduling model using a graph neural network to schedule ships.

[0044] By constructing a scheduling model using a spatiotemporal graph neural network (SPNN), the complex spatial and temporal relationships in ship scheduling can be effectively handled. Ships are treated as nodes in a graph, and the distances between ships are considered as edges. The SPNN enables deep learning and analysis of node features (such as navigation data and scheduling instructions) and edge features (such as distances between ships). In this model, a spatial aggregation function aggregates relevant information from nodes spatially, a temporal evolution function considers the evolution of ships over time, a spatial attention function focuses on and optimizes relationships between key ships, and an instruction output function generates specific scheduling instructions. Through these functions, the SPNN efficiently captures the dynamic interactions and spatiotemporal changes between ships, optimizing ship scheduling and improving the accuracy and efficiency of the scheduling process. For multi-node spatiotemporal synchronous scheduling methods supporting distributed collaborative simulation, the SPNN coordinates the computation and decision-making of each node, ensuring synchronous scheduling among multiple nodes, guaranteeing overall system collaboration and scheduling accuracy, and thus achieving a more efficient and intelligent scheduling scheme.

[0045] In this embodiment, the scheduling input dataset further includes node features and edge features. Node characteristics include driving data and scheduling instructions; The edge feature is the distance between ships; The construction of a scheduling model includes scheduling model establishment and scheduling model training; The scheduling model establishment includes spatial aggregation function, temporal evolution function, spatial attention function, and instruction output function; The process of training the scheduling model involves using historical scheduling instructions and driving data as inputs to the model, and using sailing speed and position as outputs to schedule the model.

[0046] The specific expression of the scheduling model is as follows: Spatial aggregation functions: in, In order to be in The spatial aggregation of hidden states at any given time. The spatial aggregation change weight matrix, For nodes exist The hidden state at all times For nodes exist The hidden state at all times For edge feature encoding function, In order to be in Ships at all times With ships The distance between them For aggregate functions, It is a positive integer. ; Time evolution function: in, For gated loop unit, For nodes exist The set of state changes at any given moment; Spatial attention function: in, Spatial attention coefficient, This is the transpose of the attention weight vector. For transpose operation, For the ship attention weight matrix, This is the attention weight matrix for adjacent ships. The edge feature weight matrix; Instruction output function: in, This refers to the output scheduling instructions.

[0047] Step 5: Based on the number of ships in the simulation data, create the same number of simulated ships. Input the scheduling instructions at the same time as the simulation data into the ship navigation simulation model for simulation. The ship navigation simulation model outputs the sailing speed and position of the simulated ships as the input of the scheduling model to schedule the simulated ships.

[0048] In this embodiment, step 4 specifically involves updating the simulation data at the same time interval, establishing the same number of simulated ships based on the number of ships in the generated simulation data, scheduling all simulated ships at the same time, inputting the updated simulation data and scheduling instructions into the ship navigation simulation model, outputting the sailing speed and position of the simulated ships, using the output sailing speed and position of the simulated ships as input to the scheduling model, scheduling all simulated ships, and issuing scheduling instructions to the ship navigation simulation model to complete one simulation scheduling process.

[0049] In the simulation environment, simulation data is dynamically updated at fixed time intervals, and simulated ships corresponding to the real scenario are built based on the simulation data, thereby realizing the dynamic verification and optimization of the scheduling strategy under actual operating conditions. By synchronously inputting the latest simulation data and scheduling instructions to all simulated ships at each moment, the actual sailing speed and position information of the ships are obtained and used as feedback input to the scheduling model to realize a closed-loop scheduling control process. This approach not only enhances the adaptability of the scheduling model to environmental changes, but also enables the synchronous control of multiple simulation nodes in a distributed system, ensuring that all simulated ships are scheduled under the same spatiotemporal reference, thereby improving the overall system's collaborative efficiency and scheduling accuracy. Therefore, step 4 plays a core role in the multi-node spatiotemporal synchronous scheduling method that supports distributed collaborative simulation, ensuring the timeliness, accuracy, and consistency of the scheduling process, and is an important foundation for realizing a highly reliable intelligent scheduling system.

[0050] Step 6: Based on the sailing speed, sailing time, sailing distance, and ship position of the simulated ships after scheduling by the scheduling model, calculate the comprehensive evaluation value of the scheduling area by judging the congestion coefficient of each simulated ship, and optimize the scheduling model parameters of the scheduling model according to the comprehensive evaluation value to perform real-time scheduling of ships after comprehensive evaluation.

[0051] In this embodiment, the congestion coefficient is calculated as follows: Based on the simulated ships' speed and the distance between them, the congestion coefficient for each simulated ship is calculated using the following formula: in, For the first Congestion coefficient of each ship, For ships With ships The distance between them Ships With ships sailing speed, For ships Number of ships in the vicinity.

[0052] In this embodiment, the calculation steps for the comprehensive evaluation value are as follows: By dividing the route into equal intervals, segments are formed based on these equal intervals. The number of ships in each segment is counted, with the segment as the horizontal axis and the number of ships as the vertical axis. The segment where the peak number of ships in each segment is located in the positive direction of the horizontal axis is identified as a ship congestion area. A comprehensive evaluation value is calculated by calculating the number of ships and the congestion coefficient between each area. The method for calculating the comprehensive evaluation value is as follows: in, For comprehensive evaluation, For the first in the congested area Congestion coefficient of each ship, The number of ships in the congested area. For the first A congested area The number of congested areas. This represents the average actual arrival time. The planned arrival time, This represents the average distance actually traveled. This represents the average distance the plan is to travel.

[0053] The exponent part of the formula A dynamic efficiency evaluation dimension was introduced, in which, The deviation measures time efficiency; a ratio greater than 1 indicates a time delay. The deviation in path efficiency is measured; a ratio greater than 1 indicates a longer route. Multiplying these two ratios and adding 1 ensures the exponent is always ≥1, preventing the exponent calculation from failing when the ratio is extremely small. This constitutes an "efficiency loss factor," which comprehensively captures the performance loss of the scheduling system in both spatiotemporal dimensions—the more severe the efficiency loss (i.e., the actual time and distance spent far exceed the plan), the larger the exponent value, representing the base of static congestion intensity. With the index representing dynamic efficiency loss The mathematical significance of exponentiation lies in the nonlinear amplification of the static congestion intensity. The amplification factor is determined by the degree of efficiency loss. If the scheduling is appropriate ( The exponent is 1+1=2. The overall assessment value is roughly the square of the average congestion intensity, which is a mild penalty; however, if scheduling failures lead to severe inefficiency... >> 1, the exponent becomes very large), even if the average congestion coefficient is the same, the final The value will also be amplified exponentially, thus significantly lowering the overall score. This nonlinear response ensures that the evaluation function can keenly punish scheduling schemes that, although they do not cause serious static congestion, lead to low ship navigation efficiency. This guides the scheduling model parameter optimization process not only to reduce interference between ships, but also to improve overall shipping efficiency.

[0054] The congestion coefficient of simulated ships is derived from a microscopic perspective. By analyzing the speed of each ship and its relative distance to surrounding ships, it reflects the degree of congestion in the current navigation environment. This indicator is used to assess whether an individual ship is operating in a high-risk or low-efficiency zone, providing a fine-grained reference for subsequent scheduling.

[0055] The comprehensive evaluation value is calculated from a macro perspective, performing spatial statistical analysis of vessel distribution across the entire dispatch area. By dividing the route into fixed segments, counting the number of vessels in each segment, and identifying vessel aggregation areas (such as peak traffic areas), and further combining this with the congestion situation in each segment, a global congestion assessment index is derived. This comprehensive evaluation value reflects the overall dispatch efficiency and safety status of the region, guiding the optimization of dispatch model parameters and improving overall dispatch performance.

[0056] In summary, the simulated ship congestion coefficient is used to characterize the individual operational status, while the comprehensive evaluation value is used to characterize the overall operational situation. Together, they constitute a feedback mechanism from "point" to "surface." This mechanism is of great significance to multi-node spatiotemporal synchronous scheduling methods: it can achieve rapid response to local anomalies while ensuring that the overall system remains efficient and stable throughout the collaborative simulation process, effectively supporting dynamic, intelligent, and precise scheduling in multi-node environments.

[0057] The scheduling model parameters are optimized based on the comprehensive evaluation value. These parameters include hidden state dimensions, attention dimensions, and learning rate, ensuring an optimal balance between ship efficiency, congestion, and safety during navigation. The comprehensive evaluation value, considering ship speed, time, position, and congestion coefficient, comprehensively reflects the effectiveness of the current scheduling strategy. Optimizing the scheduling model parameters can further reduce congestion, improve transportation efficiency, and thus guarantee scheduling effectiveness under different scenarios.

[0058] The Sparrow Search algorithm is used for optimization because of its powerful global search capability, good balance between exploration and utilization, and adaptability to complex systems. By simulating the foraging behavior of sparrows, the Sparrow Search algorithm can find optimal solutions within a large search space and is less prone to getting trapped in local optima. It can quickly respond to and adjust the ship scheduling model parameters in real-time scheduling, ensuring that the scheduling model can effectively cope with dynamic changes, avoid congestion, and improve the overall operating efficiency of the system. Therefore, when optimizing scheduling model parameters, the Sparrow Search algorithm can not only find suitable solutions but also handle complex scheduling tasks in a short time, meeting real-time requirements.

[0059] In this embodiment, the method for optimizing the scheduling model parameters based on the comprehensive evaluation value is to perform optimization through the sparrow search algorithm, specifically including the discoverer and the follower; The formula for updating the discoverer's location is: in, For the updated number The middle generation Only sparrows in the first The position of the sparrow To optimize the types of weights used in the optimization model within the data, the dimension... To optimize the weight dimensions in the model running on the data, For random numbers that follow a normal distribution, A uniformly random number in [0,1]. It is the warning threshold, and its value ranges from [0.5, 1]. Follower position update formula; in, The current iteration number The sparrow with the best fitness in the population was in the 1st month. The position of the dimension To generate a random number from -1 to 1, .

[0060] All the above formulas use dimensionless numerical values ​​for calculation, and the numerical values ​​substituted into the formulas are all in the International System of Units (SI). The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0061] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0062] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0063] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation, characterized in that, The specific steps include: Step 1: Obtain the historical navigation data, environmental data, and route data of each ship in the area to be dispatched, and stamp them with timestamps. Calculate the historical navigation impact coefficient of each ship at the same moment. Step 2: Construct a ship navigation simulation model. Use different ship types, navigation influence coefficients, route data, origin and destination, and dispatch instructions from historical data as inputs to the ship navigation simulation model, and the ship's speed and position as outputs to train the ship navigation simulation model. Step 3: Obtain the historical distribution probability of each factor type based on the ship type, navigation influence coefficient, and origin and destination from the historical data. Calculate the simulation occurrence probability of each factor type by fusing the historical distribution probability with random numbers. Generate simulation data for each factor type based on the simulation occurrence probability and input it into the trained ship navigation simulation model. Output the simulated ship's navigation speed and position. Step 4: Construct a scheduling input dataset with ships as nodes and the distance between ships as edges. Based on the impact of scheduling instructions on ship navigation, construct a scheduling model using a graph neural network to schedule ships. Step 5: Based on the number of ships in the simulation data, create the same number of simulated ships. Input the scheduling instructions that are sent at the same time as the simulation data into the ship navigation simulation model for simulation. The ship navigation simulation model outputs the sailing speed and position of the simulated ships as the input of the scheduling model to schedule the simulated ships. Step 6: Based on the sailing speed, sailing time, sailing distance, and ship position of the simulated ships after scheduling by the scheduling model, calculate the comprehensive evaluation value of the scheduling area by judging the congestion coefficient of each simulated ship, optimize the scheduling model parameters based on the comprehensive evaluation value, and perform real-time scheduling of ships after the comprehensive evaluation is met.

2. The multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 1, characterized in that: The environmental data includes wind speed, wind direction, water flow speed, and water flow direction; The route data includes ship traffic density and sailing intervals; The navigation data includes ship parameters, speed, direction of travel, and position. Ship parameters include tonnage and power; The specific steps for calculating the historical navigation influence coefficient of each ship at the same moment are as follows: The mass-energy product of a ship is determined based on its tonnage and power. The ship's parameter factors are determined by weighting the mass-energy product and its speed. By using environmental data, specifically the angle between wind speed, wind direction and the ship's direction of travel, we can determine the disturbance of wind on the ship's travel; by using the angle between water flow speed, water flow direction and the ship's direction of travel, we can determine the disturbance of water flow on the ship's travel; and by assigning weights to the disturbances of wind and water flow, we can calculate the environmental disturbance factor. By using route data, specifically the relationship between the angle and distance between other vessels and the route, combined with the impact of spatial attenuation on vessels, and considering the impact of route density on navigation, the route congestion factor is calculated. The navigation impact coefficient is calculated by comprehensively considering ship parameter factors, environmental disturbance factors, and route congestion factors, and by introducing nonlinear coupling effects. The specific calculation formula is as follows: in, For ship parameter factors, As environmental disturbance factors, For route congestion factor, Calculated as a navigation impact factor, Coupling index, This is a correction item for the ship's anti-interference capability.

3. The multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 1, characterized in that: The ship navigation simulation model includes an input state vector, a strategy function, a ship state update function, and a motion vector; Specifically, the ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions are used to construct the state vector of the simulation model as the input of the model, while the navigation speed and direction are used to construct the action vector as the output of the model. The ship state update function includes a position update function and a velocity update function; The system integrates multi-dimensional data such as ship type, navigation impact coefficient, route data, origin and destination, and dispatch instructions into a comprehensive state vector. This state vector serves as the basis for the ship navigation simulation model. The real-time input provides a comprehensive picture of the ship's current environment and mission status; The policy function makes decisions based on the state vector. It calculates the probability distribution of all possible action vectors in the current state through the embedded probability distribution function, and samples or selects the optimal action accordingly. After the motion vector is output, the ship state update function begins to work, updating the ship state according to the laws of physical motion: The position update function uses the current velocity, direction, and navigation influence coefficient to calculate the position at the next moment, while the velocity update function calculates the new velocity based on the acceleration command output by the strategy function, the current direction, and the velocity decay factor, thus completing the advancement of one simulation step.

4. The multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 1, characterized in that: The steps for calculating the simulation occurrence probability of each factor type by fusing historical distribution probabilities with random numbers are as follows: The method for obtaining the historical distribution probability of ship type and quantity is as follows: Within the dispatch area, by time interval The number of ships appearing within the historical time frame is statistically analyzed by type, and the probability of each type of ship appearing is calculated. The method for obtaining the historical distribution probability of the navigation impact coefficient is as follows: Based on the same time interval Calculate the navigation impact coefficient for ships within the area to be dispatched, based on the historical time length. * Within this, the navigation impact coefficient for each ship is calculated, along with the probability of each navigation impact coefficient value occurring. The method for obtaining the historical distribution probability of the starting and destination locations is as follows: Within the area to be dispatched, by time interval The statistics of the starting positions of occurrences within the historical time frame are compiled, and the probability of occurrence at each starting position is calculated.

5. A multi-node spatiotemporal synchronization scheduling method for supporting distributed collaborative simulation according to claim 1, characterized in that: The specific calculation formula for calculating the simulation occurrence probability of each factor type by fusing historical distribution probability with random numbers is as follows: in, For the first The probability of occurrence of each factor type in the simulation. For the first Historical distribution probability of each factor type The maximum random number, The random fluctuation factor follows a uniform distribution with a distribution range of [value missing]. .

6. The multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 1, characterized in that: The scheduling input dataset also includes node features and edge features. Node characteristics include driving data and scheduling instructions; The edge feature is the distance between ships; The construction of a scheduling model includes scheduling model establishment and scheduling model training; The scheduling model establishment includes spatial aggregation function, temporal evolution function, spatial attention function, and instruction output function; The process of training the scheduling model involves using historical scheduling instructions and driving data as inputs to the model, and using sailing speed and position as outputs to schedule the model.

7. A multi-node spatiotemporal synchronization scheduling method for supporting distributed collaborative simulation according to claim 1, characterized in that: The specific steps of step 4 are as follows: update the simulation data according to the same time interval, establish the same number of simulated ships according to the number of ships in the generated simulation data, schedule all simulated ships at the same time, input the updated simulation data and scheduling instructions into the ship driving simulation model, output the sailing speed and position of the simulated ships, use the sailing speed and position of the simulated ships as input to the scheduling model, schedule all simulated ships, and issue scheduling instructions to the ship driving simulation model to complete one simulation scheduling process.

8. A multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 1, characterized in that: The congestion coefficient is calculated as follows: Based on the simulated ships' speed and the distance between them, the congestion coefficient for each simulated ship is calculated using the following formula: in, For the first Congestion coefficient of an individual ship For ships With ships The distance between them Ships With ships sailing speed, For ships Number of ships in the vicinity.

9. A multi-node spatiotemporal synchronization scheduling method supporting distributed collaborative simulation according to claim 8, characterized in that: The calculation steps for the comprehensive evaluation value are as follows: By dividing the route into equal intervals, segments are formed based on these equal intervals. The number of ships in each segment is counted, with the segment as the horizontal axis and the number of ships as the vertical axis. The segment where the peak number of ships in each segment is located in the positive direction of the horizontal axis is identified as a ship congestion area. A comprehensive evaluation value is calculated by calculating the number of ships and the congestion coefficient between each area. The method for calculating the comprehensive evaluation value is as follows: in, For comprehensive evaluation, For the first in the congested area Congestion coefficient of an individual ship The number of ships in the congested area. For the first A congested area The number of congested areas. This represents the average actual arrival time. The planned arrival time, This represents the average distance actually traveled. This represents the average distance the plan is to travel.

10. A multi-node spatiotemporal synchronization scheduling method for supporting distributed collaborative simulation according to claim 1, characterized in that: The method for optimizing the scheduling model parameters based on the comprehensive evaluation value is to use a sparrow search algorithm for optimization, specifically including the discoverer and follower processes.

Images