Method for evaluating traffic volumes based on shipping databases

Abstract

The application provides a communication traffic volume evaluation method based on a shipping database, data cleaning is performed on original AIS data, and a ship space-time distribution data set is obtained; based on Starlink orbit parameters, full constellation ephemeris data sets are formed through orbit mechanics recursion, space-time registration is performed with the ship space-time distribution data set, and over-the-top satellite determination and ephemeris extraction are completed based on earth occlusion, minimum elevation angle and maximum distance constraints; in a preset time interval, a dynamic topology is sliced into a static graph, and the shortest path and performance parameters are solved through the Dijkstra algorithm; an evaluation view is constructed by comprehensively using multi-source data, an uncoupling flow aggregation model is used to recursively calculate the end-to-end performance parameters, an evaluation report is generated in multiple dimensions, and is persisted to the database; based on the evaluation results, the expected bandwidth of each ship is extracted, a multi-objective fitness function is constructed, and the bandwidth resource allocation optimization is solved through the simulated annealing algorithm.

Description

Technical Field

[0001] This invention patent relates to the field of shipping communication technology, and in particular to a method for evaluating communication traffic volume based on a shipping database. Background Technology

[0002] In recent years, marine activities such as marine tourism, nearshore aquaculture, and marine mineral exploration have developed rapidly. With the continuous increase in the number of ships, offshore platforms, and buoys, the demand for high-speed, ultra-reliable maritime communications connecting them is constantly growing. For example, the safe navigation of all ships requires navigation information and operational data, and passengers, crew, and fishermen on board need multimedia communication services. Similarly, offshore drilling platforms require real-time operational data communication, and buoys also need to upload a large amount of meteorological and hydrological data. For maritime rescue, in addition to exchanging information using text and voice, real-time video is usually required to better achieve ship-to-ship and ship-to-shore coordination. Therefore, accurate assessment of maritime communication traffic volume is of great significance for maritime transport, safe production, and emergency rescue.

[0003] The data generated by the Automatic Identification System (AIS) is characterized by its massive volume, real-time nature, and spatiotemporal complexity. Not only does it collect a large amount of information, but it also includes multi-dimensional spatiotemporal information such as the ship's dynamic position, speed, heading, route planning, and ship type. This poses a significant challenge to the storage and management of long-term historical data and the integration and sharing of multi-source information.

[0004] Traditional methods for assessing communication traffic volume based on AIS data have significant limitations: on the one hand, relying on manual statistics or simple time series models makes it difficult to handle the dynamic changes in shipping data (such as adjustments to ship routes and sudden changes in navigation status), and even more difficult to effectively uncover the spatiotemporal correlations hidden in the data; on the other hand, existing methods often focus on single-dimensional data (such as historical communication traffic), ignoring key information in AIS data such as ship type (such as differences in communication needs between cargo ships and fishing vessels) and route characteristics (such as differences in near-shore and ocean-going communication scenarios), resulting in insufficient assessment accuracy, which in turn leads to problems such as unreasonable allocation of satellite communication resources (such as bandwidth waste and congestion risks) and delayed responses to environmental interference (such as communication quality fluctuations under severe sea conditions).

[0005] To address the aforementioned challenges, database technology, with its high real-time performance, high throughput, and multi-source data fusion capabilities, has become a crucial support in the maritime communication field. By building an AIS historical database, structured storage and efficient retrieval of massive historical data can be achieved, laying the foundation for long-term data tracing and trend analysis. Furthermore, by constructing a modular maritime communication database, ship dynamic data and communication traffic data can be categorized and integrated, reducing data redundancy while providing a unified interface for the correlation analysis of multi-source information. Ultimately, this provides data-layer support for improving the accuracy of communication traffic volume assessment and optimizing resource scheduling. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, the purpose of this invention is to provide a communication traffic assessment method based on a shipping database, which solves the problems faced by data storage and communication traffic assessment under different maritime communication scenarios in existing technologies.

[0007] To address the above problems, this application proposes a communication traffic volume assessment method based on a shipping database, the method comprising:

[0008] Step 1: Obtain the basic data sources for assessing communication traffic volume, including raw AIS data and satellite data;

[0009] Step 2: Preprocess the data described in Step 1 to obtain the cleaned AIS data and form a ship spatiotemporal distribution dataset;

[0010] Step 3: Based on the aforementioned ship spatiotemporal distribution dataset, Starlink orbital parameters are recursively derived using orbital mechanics to form a full constellation ephemeris dataset; the ship spatiotemporal distribution dataset and the full constellation ephemeris dataset are spatially registered through coordinate transformation and timestamp alignment; based on the criteria of Earth occlusion removal, minimum elevation angle filtering, and maximum distance constraint, real-time determination and ephemeris data extraction of overpassing satellites are completed to obtain overpassing satellite ephemeris data;

[0011] Step 4: Based on the current satellite and ship positions, continuously dynamic topology slices are converted into static graphs within a preset time interval. The shortest path is solved using Dijkstra's algorithm, and the end-to-end feasible links and their latency and throughput performance parameters are output.

[0012] Step 5: After comprehensively cleaning the AIS data, over-the-top satellite ephemeris, end-to-end routing table, and user-preset service models, perform spatiotemporal alignment and associated binding component evaluation view. Adopt a service-decoupled traffic aggregation model to abstract services into data flows. Extract hop-by-hop path information based on the end-to-end routing table and recursively calculate end-to-end performance parameters. Generate a communication service volume evaluation report and structured evaluation results stored in the database by comprehensively considering scenario, ship, satellite, and time dimensions.

[0013] Step 6: Based on the communication traffic assessment results, extract the expected bandwidth of each ship in the current scenario, set decision variables with the upper limit of the total bandwidth of a single satellite and the minimum bandwidth satisfaction ratio as constraints, construct a multi-objective fitness function that considers priority-weighted satisfaction, Gini coefficient fairness measurement and constraint violation penalty, and use simulated annealing algorithm to solve the multi-objective constraint optimization to simulate the communication resource allocation in the current scenario.

[0014] Furthermore, the preprocessing of the data described in step one in step two is specifically carried out through the following operations: in the preprocessing stage, regional filtering is performed using rectangular geographic coordinate ranges, entries with irrelevant static attributes and missing key fields are removed, and sparse ship data containing single records are removed after sorting by time.

[0015] Independent trajectory files are generated based on the ship's MMSI, and long trajectories are segmented into navigation instances using a preset time interval as a threshold. Multi-phase data augmentation is performed on continuous trajectories, and multiple time-off trajectories are generated with the time interval as the sampling step size to expand the sample size.

[0016] During the cleaning phase, illegal tracks are identified and removed. For missing data, a classification interpolation method is used. Specifically, linear interpolation is used to fill in missing data for short time intervals, and Lagrange interpolation is used to fill in missing data for long time intervals. After filling in the missing data, the rationality is verified by combining the range of changes in ship speed and heading, and abnormal track drift values ​​are removed.

[0017] Furthermore, the calculation method for the over-the-top satellite ephemeris data mentioned in step three is as follows:

[0018] Based on the Starlink multi-level orbital constellation configuration, constellation parameters are configured to construct a three-dimensional spatial distribution model of satellites;

[0019] Based on the Kepler orbital six roots and combined with the Earth's gravitational field model, the instantaneous position and velocity of the satellite in the geocentric inertial coordinate system are calculated through the coordinate transformation matrix, and the orbital period is calculated based on the semi-major axis of the orbit. Ephemeris data containing timestamps, position and velocity are generated periodically.

[0020] The instantaneous true anomaly angle is solved by an iterative calculation process of mean anomaly angle, deviated anomaly angle, and true anomaly angle. The average angular velocity is determined by the semi-major axis in a near-circular orbit, and the angular velocity is corrected by the perturbation of Earth's oblateness. The mean anomaly angle is then recursively calculated.

[0021] The orbital correction parameters are transmitted in real time via inter-satellite links to compensate for satellite position deviations caused by orbital perturbations. These correction parameters are jointly provided by inter-satellite ranging and ground observation stations.

[0022] The system periodically calls the ephemeris calculation function to perform accuracy verification, comparing the current satellite position with the previous time. If the deviation exceeds a preset threshold, the system triggers an ephemeris update to ensure that the ephemeris data error is less than the preset accuracy.

[0023] Furthermore, the end-to-end link planning method described in step four is as follows:

[0024] The continuous time axis is discretized into continuous snapshot intervals of a preset time window. The satellite position and inter-satellite link status are frozen within each snapshot. The connection relationship structure formed between satellites through inter-satellite links is modeled as a static weighted graph.

[0025] The instantaneous spatial coordinates of satellites are calculated based on ephemeris data, and the status of inter-satellite links is determined. Adjacent satellites in the same orbital plane are determined to have sustainable and usable links based on orbital parameter matching degree and relative motion stability, and are marked as unusable during orbital machine or attitude adjustment time. Satellites in different orbital planes are determined to have link validity based on distance constraints and visibility constraints, and a network topology adjacency table for the current snapshot is generated.

[0026] Based on the static weighted graph, using hop count and propagation delay as metrics, the shortest path algorithm is used to calculate the shortest path for all nodes in the network, generating a global routing table and guiding the forwarding of data packets within the current snapshot period.

[0027] Furthermore, a transition mechanism is set at the snapshot switching boundary to ensure continuous data transmission through parallel operation of the old and new routing tables and traffic balancing scheduling.

[0028] Furthermore, the communication traffic simulation model described in step five is as follows:

[0029]

[0030] in, For the first The amount of data per communication for this type of service. The communication frequency per hour. The number of devices or users participating in the service. The number of service types; the latency simulation content includes queuing latency, transmission latency and propagation latency, and the propagation latency under the same wave position is approximated by round-trip latency.

[0031] Furthermore, the communication resource allocation simulation described in step six specifically employs a simulated annealing algorithm, including:

[0032] Population initialization, generating neighborhood solutions for each ship's bandwidth allocation scheme; fitness function.

[0033]

[0034] In the formula: To satisfy the revenue items for the business, among which The total number of ships with bandwidth to be allocated. For the first The priority weight of communication services for each ship. For the first The total revenue from overall business satisfaction is obtained by weighted summation of the ship's business demand satisfaction levels. To allocate fairness penalty items: The Gini coefficient for bandwidth allocation (characterizing the degree of uneven allocation). This is a fairness penalty coefficient; the more uneven the distribution, the greater the fitness of this deduction. Minimum bandwidth guarantee penalty This is the minimum ratio of the actual bandwidth of all vessels to the statutory / pre-set minimum guaranteed bandwidth. This is the penalty coefficient for failing to meet the basic bandwidth standard; the greater the gap in the ship's basic bandwidth, the stronger the penalty.

[0035] Acceptance criteria for calculating the objective function difference between the new solution and the current solution ;

[0036] like If so, then accept the new solution as the current solution;

[0037] like Then accept the new solution with probability.

[0038] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0039] 1. High data access efficiency: This invention employs a two-layer optimization in the AIS data storage stage, constructing a (timestamp, mmsi, data_type) composite index, which connects with the processing logic of sorting by time and separating by MMSI as described in this invention, transforming full table scans into direct index addressing; region filtering, trajectory segmentation, data augmentation, cleaning and verification are encapsulated as pre-compiled stored procedures, with the complete preprocessing pipeline executed by the database, reducing network round trips between the application and the database; enabling the entire process of AIS data from ingestion to cleaning and output to be completed within the database.

[0040] 2. The data preprocessing method provided by this invention addresses AIS data quality issues by employing a classification interpolation and rationality verification cleaning strategy. Specifically, it includes: using linear interpolation to fill in missing data over short time intervals and using Lagrange interpolation to fill in missing data over longer time intervals; performing rationality verification after interpolation in conjunction with ship navigation parameters to remove outliers. Compared to traditional cleaning methods that only use single interpolation or have no verification, this significantly improves the authenticity and usability of AIS data.

[0041] 3. The present invention has high evaluation accuracy. Its communication traffic evaluation method combines Starlink constellation precise ephemeris data, uses an improved Dijkstra algorithm to calculate the shortest path, and constructs a throughput and latency calculation model based on multiple factors. Compared with traditional evaluation methods that rely on single data or simple models, it can more comprehensively consider key factors such as ship type, route characteristics, and satellite dynamics, and the evaluation accuracy is greatly improved.

[0042] 4. This invention uses simulated annealing algorithm and dynamic bandwidth adjustment mechanism to allocate bandwidth, comprehensively considering the priority of ship services, demand satisfaction and fairness. It optimizes the allocation scheme through fitness function and can sense changes in the scenario in real time (such as the addition or removal of ships, service switching) and dynamically adjust the bandwidth. Compared with the traditional fixed or simple allocation method, it realizes efficient use of bandwidth resources, reduces the risk of congestion, and ensures the needs of high-priority services.

[0043] 5. This invention is applicable to various maritime communication scenarios such as near-shore transportation, ocean voyage, maritime rescue, and offshore drilling platform operations. It can be flexibly configured and evaluated according to the environmental parameters and communication needs of different scenarios, providing support for the optimized configuration of communication resources in different scenarios. Its scope of application far exceeds that of traditional evaluation methods for single scenarios. Attached Figure Description

[0044] Figure 1 This is a diagram of the historical database table structure as described in an embodiment of the present invention;

[0045] Figure 2 This is a diagram of the communication database table structure as described in an embodiment of the present invention;

[0046] Figure 3 This is a structural diagram of the communication information table as described in an embodiment of the present invention;

[0047] Figure 4 This is a diagram illustrating the communication service requirements in different scenarios according to embodiments of the present invention;

[0048] Figure 5 This is a flowchart illustrating the communication service simulation process of the present invention. Detailed Implementation

[0049] This invention provides a method for fusing raw AIS data, Starlink satellite ephemeris data, and ship trajectory data, and then using orbital dynamics calculations, time snapshot topology modeling, and simulated annealing optimization to assess communication traffic and dynamically allocate satellite communication resources. By applying dynamic routing planning, multi-objective traffic simulation, and resource optimization to the study of shipping satellite communication networks, the utilization rate and allocation fairness of communication resources in highly dynamic ship scenarios are significantly improved. The communication traffic assessment method based on a shipping database proposed in this invention can predict end-to-end communication traffic demands under different maritime business scenarios based on the real-time position and trajectory characteristics of ships in two-dimensional sea space, combined with the instantaneous orbital characteristics and inter-satellite link connection characteristics of low-Earth orbit satellite constellations in three-dimensional space, and generate optimal communication paths and dynamic bandwidth allocation schemes.

[0050] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0051] like Figure 5 As shown, this embodiment provides a method for evaluating communication traffic volume based on a shipping database, which mainly includes the following steps:

[0052] S101. Import raw AIS data to obtain the basic data source for communication traffic assessment. Obtain the raw NMEA format message from the AIS receiver and obtain the two-line orbital elements (TLE) of Starlink satellite as satellite data input.

[0053] S102. Preprocess the AIS data described in step one; due to the regional redundancy, mixed attributes, sparse records, and broken trajectories in the original AIS data, perform regional filtering, attribute simplification, trajectory segmentation, and data augmentation on the AIS data. Delete redundant data outside the specified area of ​​the scenario and static attributes unrelated to the ship's dynamic behavior, and remove illegal records with missing key fields and single sparse entries; segment continuous trajectories of the same MMSI into independent navigation instances according to a time threshold, and perform multi-phase time offset sampling on the effective trajectories to generate enhanced trajectories, thereby achieving data standardization and sample size expansion. In this embodiment, the data preprocessing stage specifically includes:

[0054] Preprocessing historical AIS data,

[0055] 21) Data separation: Only a specified area of ​​the scene is selected for data analysis. In this embodiment, the area is assumed to be a rectangular area, described by four geographic coordinate points, namely the lower left corner, upper left corner, upper right corner, and lower right corner of the rectangular area; AIS data of a local area is selected to reduce system memory overhead.

[0056] 22) Remove irrelevant attributes and delete unnecessary records, such as static information that is not related to the general behavior of the vessel, such as VesselName, IMO, and CallSign. Also delete attributes that are missing from the entry verification, such as the Width, Draft, Cargo, and Status columns.

[0057] 23) Sort the processed data by time;

[0058] 24) Remove entries with obviously insufficient operational records, such as ship record entries that correspond to only one record in the MMSI;

[0059] 25) Separate the ship trajectory data and generate a CSV file containing only the data of that specific ship for each ship;

[0060] 26) Based on the generated individual trajectory files, even within a given area, one MMSI may correspond to multiple flight instances, thus requiring further segmentation of the trajectory files. This preprocessing step further divides the trajectory file corresponding to one MMSI into multiple trajectory files. When the time difference between consecutive AIS data exceeds 30 minutes, the data is processed as two different trajectories.

[0061] 27) Trajectory Data Augmentation: This step augments the trajectory data of a continuous trajectory and distributes it across multiple trajectories sampled at 30-minute intervals. For example, for a 2-hour trajectory from 12:00:00 to 14:00:00, with AIS messages spaced 1-2 minutes apart, this step generates 15 trajectories from this single trajectory. The first trajectory's timestamps are 12:00:00, 12:30:00, 13:00:00, 13:30:00, and 14:00:00. The second trajectory's timestamps differ by 2 minutes, i.e., 12:02:00, 12:32:00, 13:02:00, and 13:32:00. Similarly, there will be a total of 15 trajectories. The last trajectory's timestamps differ by 28 minutes, i.e., 12:28:00, 12:58:00, 13:28:00, and 13:58:00.

[0062] This also includes cleaning historical AIS data.

[0063] AIS trajectory data is a crucial source for predicting ship trajectories. However, AIS trajectory data can contain invalid trajectories, which negatively impact subsequent modeling performance, necessitating data cleaning. This embodiment defines five types of invalid trajectory data: empty key value trajectories, duplicate trajectories, drifting trajectories, sparse trajectories, and excessively short trajectories. Empty key value trajectories are those lacking essential information; duplicate trajectories are those where multiple trajectories highly overlap in time and space; drifting trajectories indicate sudden jumps in ship trajectories; sparse trajectories show excessively sparse ship trajectory points; and excessively short trajectories represent trajectories with excessively short time intervals. Cleaning invalid trajectory data aims to ensure data quality and improve the prediction accuracy of subsequent models.

[0064] S103, The cleaning stage specifically includes:

[0065] Linear interpolation is used in this embodiment to approximate a short segment of the ship's trajectory as a straight line. A linear fit is performed on the missing point based on data from the points before and after it.

[0066] Let the location of the missing data point be... The positions of the data points before and after are and Then the interpolation formula can be expressed as:

[0067]

[0068] As shown in the above formula, linear interpolation is suitable for trajectories with relatively small curvature, but it introduces significant errors for trajectories with large curvature. This is because in regions with large curvature, linear interpolation cannot accurately approximate the true shape of the trajectory, leading to a large deviation between the interpolated results and the actual data. Therefore, when the trajectory curvature is large, higher-order interpolation methods are needed to improve the accuracy and precision of the interpolation.

[0069] Lagrange interpolation

[0070] The basic idea of ​​the Lagrange interpolation algorithm is to construct a polynomial function by fitting all known data points, and then use this function to calculate unknown data points. Assume the ship has... There are 3 known trajectory points, which respectively correspond to 1 ______ trajectory points, and they correspond to 1 ______ trajectory points, respectively. ,...., ,.in Representative ships at Longitude of time Representative ships at The latitude of the time; assuming the ship's position is different at any two times, the Lagrange interpolation polynomial can be expressed as:

[0071]

[0072] in The interpolation basis function is expressed as follows:

[0073]

[0074] Since polynomials can fit curves well, the Lagrange interpolation algorithm is more suitable for interpolating the trajectory of ships navigating curves.

[0075] See Figure 1 In the AIS data preprocessing stage described in step two, the cleaned ship spatiotemporal distribution dataset is stored in the historical database according to the following table structure:

[0076] The main table `ais_dynamic` stores vessel dynamic data. Fields include `timestamp` (UTC timestamp, corresponding to the AIS message collection time), `mmsi` (unique maritime mobility identifier for the vessel, 9 digits), `longitude`, `latitude`, `sog` (speed over land), `cog` (heading over land), and `data_type` (data type identifier, enumerated values ​​1 = dynamic data / 2 = static data / 3 = voyage data). The main table is partitioned by the `timestamp` field (e.g., by month), and a composite index is built using the three fields `(timestamp, mmsi, data_type)`. The database pre-sorts this combination to create an index tree. When users query by voyage period, specific vessel, or desired AIS data category, they can directly and quickly locate the physical storage area of ​​the data through the index, avoiding a full table scan.

[0077] The preprocessing table ais_preproc_log records the rectangular geographic coordinate range of the region selection in step two (in this embodiment, the coordinates are the four coordinate points of the lower left corner, upper left corner, upper right corner, and lower right corner), the time threshold (30 minutes) for trajectory segmentation, and the sampling step size parameters for data augmentation, which are used to trace the preprocessing process.

[0078] The cleaning rule table ais_cleanse_rule stores the judgment thresholds and interpolation strategy identifiers for five types of illegal trajectories (linear interpolation for short time intervals and Lagrange interpolation for long time intervals), and supports the parameter calls for classification interpolation and rationality verification described in step two.

[0079] For example, a coastal port generates tens of millions of AIS dynamic data entries daily. Traditionally, querying a single ship's data for a week using a single MMSI index takes 20 seconds. After adopting the three-dimensional composite index of this solution, the search scope directly locks the target time period, target ship, and dynamic data, reducing the amount of initial data screening by more than 95%. Then, by calling the predefined AIS extraction stored procedure, the steps of multiple pagination requests and temporary data filtering on the application side are eliminated, and the final query time can be compressed to less than 1 second.

[0080] S104. Generate AIS data that can be used for evaluation, forming a ship spatiotemporal distribution dataset.

[0081] S105. Load satellite data and user data. Based on the ship spatiotemporal distribution dataset, Starlink orbital parameters are recursively derived through orbital mechanics to form a full constellation ephemeris dataset. The ship spatiotemporal distribution dataset and the full constellation ephemeris dataset are spatially registered through coordinate transformation and timestamp alignment. Based on the criteria of Earth occlusion removal, minimum elevation angle screening, and maximum distance constraint, the real-time determination and ephemeris data extraction of overpassing satellites are completed to obtain overpassing satellite ephemeris data.

[0082] In this embodiment, obtaining the over-the-top satellite ephemeris data based on the preprocessed data specifically involves:

[0083] 1) Constellation parameter configuration,

[0084] The Starlink constellation employs a multi-level, layered orbital deployment configuration, consisting of tens of thousands of satellites (deployed in phases, with a final target of over 40,000 satellites), distributed across multiple orbital planes (over 100 orbital planes have been planned and are dynamically adjusted with each deployment phase). The satellites operate in low Earth orbits at various altitudes, including approximately 540 km and 1150 km, with orbital inclinations ranging from 53 degrees and 70 degrees to approximately 97.6 degrees required for polar orbits. By rationally planning the distribution of satellites within orbital planes and the phase relationships between satellites in adjacent orbital planes, an orderly and dense distribution structure is achieved in three-dimensional space. This multi-level, multi-inclination orbital design not only ensures continuous global coverage (including high latitudes and polar regions) but also optimizes the overall communication resource utilization of the system through inter-satellite collaboration and dynamic resource scheduling, meeting the high-speed broadband communication needs of a large number of users.

[0085] 2) Ephemeris Data Calculation Method

[0086] Based on the Kepler orbital parameters (semi-major axis, eccentricity, inclination, right ascension of the ascending node, argument of perigee, and true anomaly), and combined with the Earth's gravitational field model, the instantaneous position of the satellite in the geocentric inertial coordinate system (J2000) is calculated using formulas. and speed The formula is as follows:

[0087]

[0088]

[0089]

[0090]

[0091] Where r is the distance from the satellite to the Earth's center; e is the orbital eccentricity; and ν is the true anomaly angle. The radial velocity of the satellite; The satellite's tangential velocity;

[0092] Through formula Obtain the semi-major axis of the orbit Based on the formula:

[0093]

[0094] The orbital period T is calculated, and thus the periodic time of the constellation satellites is obtained.

[0095] Based on the variation of the true anomaly angle over time, the satellite position and velocity are updated every 5 seconds (synchronized with subsequent snapshot periods) to generate ephemeris data. This ephemeris data includes a timestamp, position, and velocity. The variation of the true anomaly angle ν over time is achieved through iterative calculations of the mean anomaly angle, the deviated anomaly angle, and the true anomaly angle. The formulas and steps are as follows:

[0096] In a near-circular orbit, the average angular velocity of the satellite From the semi-major axis of the track Decide:

[0097]

[0098] Introduction Perturbation correction angular velocity

[0099]

[0100] Calculate the change of the angle of approach with time.

[0101]

[0102]

[0103] in Let t be the Earth's gravitational constant, and t be the target time. As the initial reference time, Let be the true anterior angle at time t. Initial time The true anomaly angle, where e is the orbital eccentricity. This is the Earth's oblateness coefficient;

[0104] Satellite position deviations caused by orbital perturbations (atmospheric drag, lunar and solar gravity, etc.) are compensated for in real-time by orbital correction parameters transmitted via inter-satellite links. These parameters are jointly provided by inter-satellite ranging and ground observation stations, with the injection frequency matched to the satellite's orbital dynamics to ensure ephemeris data errors are less than 50 meters. The module calls the ephemeris calculation function every 10 seconds to compare the current satellite position with the previous position. If the deviation exceeds 30 meters, an ephemeris update is triggered and synchronously pushed to the communication link establishment and routing calculation module, providing real-time data support for global broadband communication and other services.

[0105] like Figure 2 As shown, in step three, the constellation configuration and ephemeris calculation stage, Starlink orbital parameters and propagation constraints are managed according to the following table structure:

[0106] The constellation layer configuration table starlink_shell records multi-layer orbital constellation configuration parameters, including shell_id (orbital layer identifier), altitude (orbital altitude, such as 540km, 1150km), inclination (orbital inclination angle, such as 53°, 70°, 97.6°), plane_count (number of orbital planes), and sat_per_plane (number of satellites per plane), which are used to construct a three-dimensional spatial distribution model of satellites.

[0107] Orbital elements table tle_ephemeris: Stores the six Keplerian orbital elements (semi-major axis a, eccentricity e, inclination i, right ascension of ascending node Ω, argument of perigee ω, true perigee ν) for each satellite, as input for the ephemeris recursion of formulas (1)-(9).

[0108] The link constraint parameter table link_constraint stores the Earth's radius R_ground (6378km), Earth's gravitational constant μ, Earth's oblateness coefficient J_2, maximum inter-satellite link distance (6200km), minimum elevation angle (4°), ephemeris update deviation threshold (30 meters), and ephemeris accuracy target (50 meters), and is used for determining over-the-head satellites and link validity.

[0109] The orbit correction table orbit_correction stores orbit correction parameters jointly provided by inter-satellite link ranging and ground observation stations. It is periodically injected to compensate for satellite position deviations caused by perturbations such as atmospheric drag and lunar gravity.

[0110] S106. Based on the current satellite position and ship position, the continuous dynamic topology is sliced ​​into a static map within a preset time interval. The shortest path is solved by Dijkstra's algorithm, and the end-to-end feasible links and their delay and throughput performance parameters are output.

[0111] In this embodiment, the satellite routing calculation specifically includes:

[0112] To address the dynamic changes in network topology caused by the high-speed movement of Starlink constellation satellites, a shortest path routing strategy based on snapshot sequences is adopted. The continuous time axis is discretized into a series of fixed-duration snapshot intervals, with each snapshot period set to 12 seconds. Within each snapshot interval, the satellite positions and inter-satellite link connection status are temporarily frozen, and the network topology is treated as a static weighted graph. This transforms the dynamic routing problem into a series of static routing computation problems, significantly reducing processing complexity.

[0113] At the start of each snapshot period, the system first calculates the instantaneous spatial coordinates of all satellites based on Starlink's high-precision ephemeris data. Then, it enters the inter-satellite link status determination phase. For adjacent satellites within the same orbital plane, due to their high orbital parameter matching and stable relative motion, they are determined to be continuously available links, only temporarily marked as unavailable when the satellite performs orbital maneuvers or attitude adjustments. For satellites in different orbital planes, both distance and visibility conditions must be met—the straight-line distance between the two satellites must not exceed 6200 km, and the link connection must not be obstructed by the Earth (minimum elevation angle ≥ 4°)—to be considered a valid link. Through this process, the network topology adjacency table for the current snapshot is quickly generated.

[0114] During the route calculation phase, based on the static topology of the current snapshot, and using hop count and propagation delay (link distance divided by the speed of light) as comprehensive metrics, an improved Dijkstra algorithm is employed to calculate the shortest path for all satellite nodes in the entire network, generating a global routing table. Traditional Dijkstra uses only a single distance weight; this scheme reconstructs the comprehensive cost function.

[0115]

[0116] in, Total number of hops on the path; Link straight-line distance; : Speed ​​of light in vacuum; , Adjustable weighting coefficients for maritime scenarios. Magnification for long-distance ocean transmission scenarios. Prioritize reducing transmission delay; adjust the intensity of transmission in densely populated nearshore areas. This reduces queuing losses caused by multi-hop forwarding; for high-priority rescue services, latency weights can be dynamically increased to ensure real-time performance.

[0117] This routing table is effective within the current 12-second snapshot period, guiding the forwarding of all data packets. When the next snapshot period boundary is reached, the system recalculates the topology and generates a new routing table, achieving a comprehensive update of routing information.

[0118] To address potential packet loss during snapshot switching, the system employs a 0.5-second transition mechanism. During this transition, both the old and new routing tables operate simultaneously, ensuring data transmission continuity through traffic balancing. Additionally, a full network topology trend prediction is performed every four snapshot cycles (48 seconds) to proactively identify potential link outage risks. Unstable links are prioritized for avoidance in subsequent snapshot calculations, further enhancing routing reliability. This mechanism is well-suited to the large-scale, highly dynamic nature of the Starlink constellation, maintaining routing accuracy while keeping computational overhead within the system's tolerance.

[0119] S108. Generate communication traffic assessment data. Integrate and spatially align and bind the cleaned AIS data, over-the-top satellite ephemeris, end-to-end routing table, and user-preset service models to form an assessment view. Employ a service-decoupled traffic aggregation model to abstract services into data flows. Extract hop-by-hop path information from the end-to-end routing table and recursively calculate end-to-end performance parameters. Generate a communication traffic assessment report by integrating scenario, ship, satellite, and time dimensions.

[0120] S109. The evaluation results are structured and stored in the database;

[0121] like Figure 4 As shown, the user-preset business model is configured according to the maritime scenario dimension, and the communication service requirement parameters under different scenarios are as follows:

[0122] In near-shore shipping scenarios, the main business types are AIS messages and voice communication. The data volume per AIS message... Approximately 256 bits, communication frequency Approximately 60 times per hour, number of devices involved per ship For 3-5 units; voice service Approximately 64kbit Approximately 10 times per hour. In this scenario, the ship density is high and the navigation trajectory is relatively regular; therefore, the priority weight wi in step S106 is set to medium.

[0123] Ocean voyage scenario: Add video backhaul service to AIS message and voice communication. Video service data volume per transmission. Approximately 2 Mbit, communication frequency Approximately 4 times per hour, number of devices per ship Increase to 8-12 units (including surveillance cameras). In this scenario, satellite coverage is sparse, and step S104 should prioritize ensuring the stability of a continuously available link.

[0124] Maritime rescue scenario: Primarily based on high-priority video and real-time voice collaboration. Video service data volume per session. Approximately 8 Mbit Based on dynamic event triggering (no fixed cycle), number of devices per ship Up to 15 units (including drone image transmission equipment). In step S106, the priority_level of the ship in this scenario is automatically set to the highest level, with a weight wi that is significantly higher than other scenarios, and an event-driven emergency adjustment mechanism is triggered.

[0125] Offshore drilling platform operation scenario: This is a fixed platform scenario, with business primarily involving sensor data reporting and operational video inspection. Sensor data Smaller (about 1kbit), but Fixed (e.g., reporting every 10 minutes); video inspection Approximately 4 Mbit Triggered by work shift. In this scenario, the ship's position is fixed, and the over-the-head satellite determination in step three can be simplified to quasi-static coverage analysis.

[0126] The above scenario parameters are passed through Figure 2 The `scene_type` field in the `vessel_profile` table is bound to the vessel's MMSI. When calculating traffic volume, the system automatically calls the appropriate field based on the current vessel's scenario. Figure 4 The corresponding scene , , Substituting parameters into the formula Perform aggregation calculations and based on Figure 2 The hop-by-hop path information in the route_table is used to recursively calculate end-to-end latency and throughput.

[0127] In this embodiment, communication traffic volume is calculated based on user behavior distribution. Specifically, calculating the communication traffic volume of a ship requires considering factors such as its service type, application requirements, communication protocol, data generation frequency, number of devices, and communication mode to construct a communication traffic model. This invention categorizes communication service types into transmission data types such as AIS messages, voice, video, and images. Therefore, communication services can be converted into the following model:

[0128]

[0129] No. The amount of data transmitted per communication for this type of service (in bits). Communication frequency per hour (unit: times / h) The number of devices or users participating in this service. Number of business types;

[0130] like Figure 3 As shown, during the evaluation view construction phase, multi-source data achieves spatiotemporal alignment and association binding through the following table structure in the communication database:

[0131] The vessel profile table stores vessel attributes associated with the cleaned AIS data, including mmsi, vessel_type (vessel type, such as cargo ship / passenger ship / fishing vessel / rescue vessel), and scene_type (scene type, associated with...). Figure 4 (In the context of nearshore / oceanic / rescue / drilling scenarios), and priority_level (communication service priority level).

[0132] The overhead satellite ephemeris table (satellite_overhead) stores the overhead satellite ephemeris extracted in step three. Its fields include satellite_id, timestamp, x, y, z (position in J2000 coordinate system), and v. x v y v z (Speed), associated with the ais_dynamic table via timestamp.

[0133] End-to-end routing table (route_table): Stores the routing results generated in step four. Fields include snapshot_id (snapshot identifier, corresponding to a 12-second cycle), source_mmsi (source vessel MMSI), dest_node (destination node identifier), path_hops (hop-by-hop satellite sequence, stored as a JSON array or string), total_delay (end-to-end total delay), and throughput (link throughput).

[0134] The assessment view is a virtual view generated by associating the three tables above through the timestamp and mmsi fields. It is used as input for the business-specific decoupled traffic aggregation model. The view fields include ship location, overhead satellite location, end-to-end path, service type, latency, and throughput, enabling a unified interface query for multi-source data.

[0135] To accurately assess the service response capability in maritime ship communications, this invention calculates end-to-end latency in three parts: processing latency, transmission latency, and propagation latency, and models each part separately. The service arrival and service process within the processing latency follows a Poisson distribution, and the processing latency parameters are set according to the service priority.

[0136] Transmission latency is calculated based on data packet size and link bandwidth, using the following model: ,in The data packet size (in bits). This is the bandwidth. The propagation delay is related to the physical distance; assuming the signal propagates at the speed of light, its propagation delay is calculated as... ,in The propagation distance is given. To improve computational efficiency, the propagation delay at the same wavelength is approximated using the round-trip time (RTT) of a single read.

[0137] Ultimately, the total end-to-end latency model is expressed as the sum of three terms: Total latency = queuing latency + transmission latency + propagation latency.

[0138] S107, during execution, firstly, the expected bandwidth of each ship is extracted from the previous evaluation results in S108, and used as the input parameters for the simulated annealing algorithm; S108, after optimization in S107, persistently associates and binds the final bandwidth allocation scheme with the evaluation data. Based on the previous communication traffic volume evaluation results, the expected bandwidth of each ship in the current scenario is extracted, and decision variables are set with the upper limit of the total bandwidth of a single satellite and the minimum bandwidth satisfaction ratio as constraints. A multi-objective fitness function considering priority-weighted satisfaction, Gini coefficient fairness measurement, and constraint violation penalties is constructed. The simulated annealing algorithm is used to solve the multi-objective constraint optimization to simulate the communication resource allocation in the current scenario.

[0139] In this embodiment, in a scenario where multiple ships share limited satellite communication bandwidth, the key to improving overall system performance lies in how to rationally allocate bandwidth resources while meeting differentiated service needs and priorities. To this end, this invention provides a bandwidth resource allocation mechanism based on simulated annealing algorithm. This mechanism optimizes the bandwidth allocation scheme under multi-objective constraints, ensuring fair and effective bandwidth resource allocation while guaranteeing priority requirements.

[0140] I. The design process of the simulated annealing algorithm is as follows:

[0141] (1) Population initialization

[0142] Initialize a population where each individual represents a possible bandwidth allocation scheme, represented as an array of length N (where N is the number of ships). Each element represents the allocated bandwidth value for that ship. The initial population satisfies basic constraints (e.g., the total bandwidth does not exceed the bandwidth limit); set the initial temperature. Temperature decay coefficient α (0 < α < 1), termination temperature .

[0143] (2) Objective function design

[0144] The objective function is used to measure the "quality" of a bandwidth allocation scheme, taking into account the following three aspects:

[0145] Fairness: The Gini coefficient measures the dispersion of resource allocation; the smaller the coefficient, the fairer the allocation.

[0146] Satisfaction: Calculated by the ratio of "actual allocated bandwidth / expected bandwidth" for each vessel;

[0147] Priority Weighting: Introducing weights based on the vessel's service level (high, medium, low);

[0148] The final fitness function takes the following form:

[0149]

[0150] in For the first The priority weight of each vessel, and the violation of constraints include exceeding the bandwidth limit, allocating less than the minimum proportion, etc.

[0151] (3) Generation of neighborhood solutions

[0152] For the current bandwidth allocation scheme, randomly select one or more ships and make small adjustments to their current bandwidth (such as increasing or decreasing a certain amount of bandwidth Δb) to generate a new bandwidth allocation scheme as a neighborhood solution.

[0153] (4) Acceptance criteria judgment

[0154] Calculate the difference in objective function between the new solution and the current solution. ;like If so, then accept the new solution as the current solution; if Then, based on probability Accept the new solution (T is the current temperature).

[0155] (5) Temperature update

[0156] The temperature is updated according to the formula T=α⋅T, so that the temperature gradually decreases.

[0157] (6) Termination conditions

[0158] When temperature The algorithm terminates when the objective function value shows no significant improvement after multiple iterations, and outputs the optimal bandwidth allocation scheme for the objective function.

[0159] The dynamic bandwidth adjustment mechanism described in this embodiment relies on AIS real-time data, satellite ephemeris, and communication traffic assessment results to achieve real-time iteration and dynamic correction of bandwidth. It is divided into three scenarios: periodic polling adjustment, event-triggered emergency adjustment, and cross-satellite handover adjustment. It adopts a simulated annealing algorithm to solve the bandwidth adaptation problem caused by ship movement, sudden changes in services, and satellite handover. The specific implementation logic is as follows:

[0160] 1. Basic periodic polling adjustment (normal scenario)

[0161] Based on the aforementioned 12-second network snapshot cycle, a regular bandwidth adjustment cycle is defined as three snapshot cycles (36 seconds). Within each cycle, the latest AIS data is first retrieved to update the ship's position, navigation status, and number of online ships. Then, real-time ephemeris data is used to determine the ship's current satellite access, link quality, and remaining bandwidth. The expected bandwidth, service priority, and traffic requirements for each ship are recalculated, and the simulated annealing algorithm is invoked to re-solve for the optimal bandwidth allocation scheme, automatically replacing the previous allocation strategy. This method is suitable for routine near-shore and ocean-going navigation scenarios where ships move slowly and service traffic changes steadily.

[0162] 2. Emergency adjustments triggered by events (unexpected scenarios)

[0163] In this embodiment, a preset trigger threshold is used. Once an abnormal event is detected, the normal cycle is immediately interrupted, and an emergency bandwidth reallocation is initiated.

[0164] Sudden changes in business traffic, with the instantaneous traffic of a single vessel exceeding the historical average by more than 30% (such as enabling high-definition video transmission or emergency video transmission at sea).

[0165] Changes in vessel status, including newly added vessels joining the network, vessels leaving the network, and high-priority operation vessels (drilling platforms, rescue vessels) entering the coverage area;

[0166] When link quality deteriorates, the elevation angle between the ship and the satellite decreases and the distance increases, resulting in increased link loss and decreased throughput. After triggering, priority is given to ensuring bandwidth for emergency and high-priority services, while bandwidth of idle ships is reduced, the fitness function is recalculated, and a new allocation scheme is quickly output to avoid communication congestion and service interruption.

[0167] 3. Cross-satellite handover adjustment (satellite replacement scenario)

[0168] Due to the high-speed movement of low-Earth orbit satellites, individual ships will periodically leave the current satellite coverage and connect to new overpass satellites. Satellite switching actions can be predicted 5 seconds in advance using ephemeris data.

[0169] During the handover transition period (using the 0.5s routing transition mechanism): the old and new satellites carry services in parallel, and the bandwidth is distributed proportionally to ensure uninterrupted transmission;

[0170] After the switchover is completed: Based on the total bandwidth limit of the new satellite and the list of connected ships, simulated annealing optimization is re-executed, and the bandwidth of the entire ship is reconfigured by combining the latency and throughput parameters of the new link; at the same time, the bandwidth allocation logs before and after the satellite switchover are recorded and stored in the shipping database to provide historical basis for the bandwidth prediction of subsequent ships on the same trajectory.

[0171] 4. Constraints and safety net guarantees

[0172] The entire dynamic adjustment process strictly adheres to two hard constraints: first, a maximum total bandwidth limit for a single satellite, meaning the sum of all allocated bandwidths must not exceed the satellite's rated bandwidth; second, a minimum bandwidth guarantee, ensuring that the actual bandwidth of all vessels in the network must not be lower than the preset minimum guaranteed bandwidth. If the optimization solution touches the constraint threshold, the algorithm automatically increases the penalty term and iterates again to solve the problem, balancing allocation flexibility with basic communication reliability.

[0173] S110, End.

[0174] In summary, compared with the traditional fixed bandwidth allocation mode, the dynamic bandwidth adjustment mechanism significantly improves the utilization rate of satellite bandwidth resources, while adapting to the highly dynamic and sudden service characteristics of maritime communications.

[0175] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0176] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating communication traffic volume based on a shipping database, characterized in that, The method includes: Step 1: Obtain the basic data sources for assessing communication traffic volume, including raw AIS data and satellite data; Step 2: Preprocess the data described in Step 1 to obtain the cleaned AIS data and form a ship spatiotemporal distribution dataset; Step 3: Based on the aforementioned ship spatiotemporal distribution dataset, Starlink orbital parameters are recursively derived using orbital mechanics to form a full constellation ephemeris dataset; the ship spatiotemporal distribution dataset and the full constellation ephemeris dataset are spatially registered through coordinate transformation and timestamp alignment; based on the criteria of Earth occlusion removal, minimum elevation angle filtering, and maximum distance constraint, real-time determination and ephemeris data extraction of overpassing satellites are completed to obtain overpassing satellite ephemeris data; Step 4: Based on the current satellite and ship positions, continuously dynamic topology slices are converted into static graphs within a preset time interval. The shortest path is solved using Dijkstra's algorithm, and the end-to-end feasible links and their latency and throughput performance parameters are output. Step 5: After comprehensively cleaning the AIS data, over-the-top satellite ephemeris, end-to-end routing table, and user-preset service models, perform spatiotemporal alignment and associated binding component evaluation view. Adopt a service-decoupled traffic aggregation model to abstract services into data flows. Extract hop-by-hop path information based on the end-to-end routing table and recursively calculate end-to-end performance parameters. Generate a communication service volume evaluation report and structured evaluation results stored in the database by comprehensively considering scenario, ship, satellite, and time dimensions. Step 6: Based on the communication traffic assessment results, extract the expected bandwidth of each ship in the current scenario, set decision variables with the upper limit of the total bandwidth of a single satellite and the minimum bandwidth satisfaction ratio as constraints, construct a multi-objective fitness function that considers priority-weighted satisfaction, Gini coefficient fairness measurement and constraint violation penalty, and use simulated annealing algorithm to solve the multi-objective constraint optimization to simulate the communication resource allocation in the current scenario.

2. The communication traffic volume assessment method based on a shipping database according to claim 1, characterized in that, Step 2 involves preprocessing the data described in Step 1. Specifically, the AIS data is preprocessed through the following operations: In the preprocessing stage, the region is filtered by rectangular geographic coordinate range, entries with irrelevant static attributes and missing key fields are removed, and sparse ship data containing single records are removed after sorting by time. Independent trajectory files are generated based on the ship's MMSI, and long trajectories are segmented into navigation instances using a preset time interval as a threshold. Multi-phase data augmentation is performed on continuous trajectories, and multiple time-off trajectories are generated with the time interval as the sampling step size to expand the sample size. During the cleaning phase, illegal tracks are identified and removed. For missing data, a classification interpolation method is used. Specifically, linear interpolation is used to fill in missing data for short time intervals, and Lagrange interpolation is used to fill in missing data for long time intervals. After filling in the missing data, the rationality is verified by combining the range of changes in ship speed and heading, and abnormal track drift values ​​are removed.

3. The communication traffic volume assessment method based on a shipping database according to claim 1, characterized in that, The calculation method for the over-the-top satellite ephemeris data mentioned in step three is as follows: Based on the Starlink multi-level orbital constellation configuration, constellation parameters are configured to construct a three-dimensional spatial distribution model of satellites; Based on the Kepler orbital six roots and combined with the Earth's gravitational field model, the instantaneous position and velocity of the satellite in the geocentric inertial coordinate system are calculated through the coordinate transformation matrix, and the orbital period is calculated based on the semi-major axis of the orbit. Ephemeris data containing timestamps, position and velocity are generated periodically. The instantaneous true anomaly angle is solved by an iterative calculation process of mean anomaly angle, deviated anomaly angle, and true anomaly angle. The average angular velocity is determined by the semi-major axis in a near-circular orbit, and the angular velocity is corrected by the perturbation of Earth's oblateness. The mean anomaly angle is then recursively calculated. The orbital correction parameters are transmitted in real time via inter-satellite links to compensate for satellite position deviations caused by orbital perturbations. These correction parameters are jointly provided by inter-satellite ranging and ground observation stations. The system periodically calls the ephemeris calculation function to perform accuracy verification, comparing the current satellite position with the previous time. If the deviation exceeds a preset threshold, the system triggers an ephemeris update to ensure that the ephemeris data error is less than the preset accuracy.

4. The communication traffic volume assessment method based on a shipping database according to claim 1, characterized in that, The end-to-end link planning method described in step four is as follows: The continuous time axis is discretized into continuous snapshot intervals of a preset time window. The satellite position and inter-satellite link status are frozen within each snapshot. The connection relationship structure formed between satellites through inter-satellite links is modeled as a static weighted graph. Calculate the instantaneous spatial coordinates of the satellite based on ephemeris data and determine the status of the inter-satellite link; Adjacent satellites in the same orbital plane are determined to have a sustainable and usable link based on orbital parameter matching degree and relative motion stability, and are marked as unusable during orbital machine or attitude adjustment time; satellites in different orbital planes determine the link validity based on distance constraints and visibility constraints, and generate a network topology adjacency table for the current snapshot; Based on the static weighted graph, using hop count and propagation delay as metrics, the shortest path algorithm is used to calculate the shortest path for all nodes in the network, generating a global routing table and guiding the forwarding of data packets within the current snapshot period.

5. The communication traffic volume assessment method based on a shipping database according to claim 4, characterized in that, A transition mechanism is set at the snapshot switching boundary to ensure continuous data transmission through parallel operation of the old and new routing tables and traffic balancing scheduling.

6. The communication traffic volume assessment method based on a shipping database according to claim 1, characterized in that, The communication traffic simulation model described in step five is as follows: ， in, For the first The amount of data per communication for this type of service. The communication frequency per hour. The number of devices or users participating in the service. The number of service types; the latency simulation content includes queuing latency, transmission latency and propagation latency, and the propagation latency under the same wave position is approximated by round-trip latency.

7. The communication traffic volume assessment method based on a shipping database according to claim 1, characterized in that, Step six, the communication resource allocation simulation, specifically employs a simulated annealing algorithm, including: Population initialization, generating a neighborhood fitness function for each ship's bandwidth allocation scheme. ， In the formula: To satisfy the revenue items for the business, among which The total number of ships with bandwidth to be allocated. For the first The priority weight of communication services for each ship. For the first The total revenue from overall business satisfaction is obtained by weighted summation of the ship's business demand satisfaction levels. To allocate fairness penalty items: The Gini coefficient for bandwidth allocation (characterizing the degree of uneven allocation). This is a fairness penalty coefficient; the more uneven the distribution, the greater the fitness of this deduction. Minimum bandwidth guarantee penalty This is the minimum ratio of the actual bandwidth of all vessels to the statutory / pre-set minimum guaranteed bandwidth. This is the penalty coefficient for failing to meet the basic bandwidth standard; the greater the gap in the ship's basic bandwidth, the stronger the penalty. Acceptance criteria for calculating the objective function difference between the new solution and the current solution ; like If so, then accept the new solution as the current solution; like If so, then accept the new solution with probability.

Images