Ship departure time prediction method and device, electronic equipment, medium and chip

Abstract

Embodiments of the present invention provide a method, apparatus, electronic device, medium, and chip for predicting ship departure times. The method includes: acquiring first correlation data of a target ship at a target port; determining at least one internal correction parameter and at least one external correction parameter of the target ship based on the first correlation data; determining the expected departure time of the target ship based on the internal and external correction parameters; acquiring actual parameters; determining an overall prediction error based on the actual departure time and the expected departure time; determining the internal parameter error; determining the external correction error; acquiring second correlation data of at least one subsequent ship at the target port; and determining the expected departure time of the subsequent ship at the target port based on the internal parameter error, the external correction error, and the second correlation data. The solution of the present invention improves the accuracy and reliability of ship departure time prediction results.

Description

Technical Field

[0001] This invention relates to the field of maritime management technology, and more specifically, to a method, apparatus, electronic device, medium, and chip for predicting ship departure time. Background Technology

[0002] The prediction of ship departure time mainly relies on the static plans released by shipping companies, which cannot effectively integrate multi-source dynamic information such as real-time port operations, weather, and abnormal events. In addition, the model lacks self-learning ability, which means that the prediction results cannot be dynamically calibrated according to the actual operational feedback. In the scenario of multiple port calls, the error is easy to accumulate step by step, resulting in insufficient accuracy and stability of the prediction. Summary of the Invention

[0003] The purpose of this invention is to provide a method, device, electronic device, medium, and chip for predicting ship departure time, which can solve the problems of insufficient accuracy and stability in predicting ship departure time.

[0004] In view of this, an embodiment of the first aspect of the present invention provides a method for predicting ship departure time.

[0005] A second aspect of the present invention provides a device for predicting ship departure time.

[0006] An embodiment of the third aspect of the present invention provides an electronic device.

[0007] An embodiment of the fourth aspect of the present invention provides a readable storage medium.

[0008] An embodiment of the fifth aspect of the present invention provides a chip.

[0009] To achieve the above objectives, an embodiment of the first aspect of the present invention provides a method for predicting ship departure time, comprising: acquiring first correlation data of a target ship at a target port; determining at least one internal correction parameter and at least one external correction parameter of the target ship based on the first correlation data; determining the expected departure time of the target ship based on the internal correction parameter and the external correction parameter; acquiring actual parameters of the target ship at the target port, the actual parameters including the actual departure time; determining an overall prediction error based on the actual departure time and the expected departure time; determining an internal parameter error based on the actual parameters and the internal correction parameter; determining an external correction error based on the overall prediction error and the internal parameter error; acquiring second correlation data of at least one subsequent ship at the target port; and determining the expected departure time of the subsequent ship at the target port based on the internal parameter error, the external correction error, and the second correlation data.

[0010] This method constructs a hierarchical error correction model, which systematically decomposes the total error of departure time prediction into internal factor errors originating from the ship's own operations and external factor errors originating from the port environment. By using error analysis of ships that have completed their operations, the prediction of subsequent ships is dynamically corrected, thereby achieving high-precision and adaptive time estimation.

[0011] Understandably, by acquiring and quantifying the internal and external factors affecting departure operations, a leap from rough estimation to accurate prediction of departure time has been achieved. The overall prediction error is decomposed into internal parameter errors originating from the ship's own operations and external correction errors originating from the port environment. This decomposition result is used to dynamically correct subsequent ship predictions, thereby constructing an intelligent prediction closed loop with self-learning and self-adaptive capabilities, which improves the accuracy and reliability of the prediction results.

[0012] In some technical solutions, optionally, the first associated data of the target vessel at the target port can be obtained, including: obtaining raw data from multiple heterogeneous data sources, including the target vessel's schedule data, the target port's operation data, meteorological and sea condition data, and abnormal event data; cleaning the raw data to determine preprocessed data; and performing spatiotemporal alignment processing on the preprocessed data to map the preprocessed data corresponding to the schedule data, port operation data, meteorological and sea condition data, and abnormal event data to a unified spatiotemporal reference system to determine the first associated data.

[0013] In this solution, in the face of the current situation in the shipping and port industry where data sources are scattered, formats are inconsistent, and spatiotemporal benchmarks vary, the original chaotic information flow is transformed into aligned, structured, and correlated data that can be directly used for quantitative analysis through preset data processing rules. This lays a reliable data foundation for subsequent accurate parameter calculations and time predictions.

[0014] In some technical solutions, optionally, determining at least one internal correction parameter and at least one external correction parameter for the target vessel based on the first associated data includes: determining the port waiting time based on the preprocessed data corresponding to the port operation data in the first associated data; determining the predicted operation time corresponding to the target vessel based on the preprocessed data corresponding to the ship schedule data and the port operation data in the first associated data; and determining the internal correction parameter based on the port waiting time and the predicted operation time.

[0015] In this approach, the cleaned and aligned high-quality primary correlation data is transformed into internal correction parameters for quantifying the time spent by vessels on core operations in port. These internal correction parameters directly reflect the time investment required for the target vessel to complete its planned operations.

[0016] The essence of internal correction parameters is to quantify and structure the time spent on key activities of ships within port, providing a solid foundation for generating an overall forecast by superimposing them with external factors. Through this decomposition and calculation, internal correction parameters are no longer fuzzy empirical estimates, but rather derivations based on real-time data and explicit rules, significantly improving the objectivity and accuracy of the forecast.

[0017] In some technical solutions, optionally, at least one internal correction parameter and at least one external correction parameter of the target vessel are determined based on the first associated data, including: determining the external correction parameter based on the preprocessed data corresponding to the meteorological and sea state data and the abnormal event data in the first associated data.

[0018] This plan clarifies that external correction parameters are primarily determined based on two types of information reflecting the port's external environment and unforeseen circumstances: meteorological and sea state data, and abnormal event data. External correction parameters are used to quantify external factors that are not directly controlled by the ship and port's operational plans but can interfere with their processes and cause additional time delays.

[0019] In some technical solutions, the actual parameters may optionally include the actual waiting time and the actual operation time. The internal parameter error is determined based on the actual parameters and the internal correction parameters, including: determining the waiting time error based on the actual waiting time and the port waiting time; and determining the operation time error based on the actual operation time and the predicted operation time.

[0020] In this scheme, after the target vessel actually departs, the internal correction parameters on which the previous prediction was based are accurately compared with the final actual operation time data, thereby quantifying the deviation of the prediction model in assessing the time consumption of the vessel's own operation process.

[0021] In some technical solutions, optionally, the estimated departure time of subsequent vessels at the target port is determined based on internal parameter errors, external correction errors, and second correlation data, including: obtaining preset weighting coefficients; determining correction parameters based on weighting coefficients, external correction parameters, and external correction errors; determining port operation efficiency parameters based on weighting coefficients, port waiting time, and actual waiting time; determining waiting time parameters based on weighting coefficients, predicted operation time, and actual operation time; and determining the estimated departure time of subsequent vessels based on correction parameters, port operation efficiency parameters, waiting time parameters, and second correlation data.

[0022] In this scheme, the prediction error knowledge obtained from the analysis of the preceding target ships will be applied to the prediction of subsequent ships through a set of quantitative correction models, thereby achieving iterative improvement in prediction accuracy.

[0023] A second aspect of the present invention provides a vessel departure time prediction device, comprising: a data acquisition module for acquiring first correlation data of a target vessel at a target port; a parameter determination module for determining at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first correlation data; a model prediction module for determining the expected departure time of the target vessel based on the internal correction parameter and the external correction parameter; an actual parameter module for acquiring actual parameters of the target vessel at the target port, the actual parameters including the actual departure time; an overall error module for determining an overall prediction error based on the actual departure time and the expected departure time; an internal error module for determining an internal parameter error based on the actual parameters and the internal correction parameter; an external error module for determining an external correction error based on the overall prediction error and the internal parameter error; a correlation acquisition module for acquiring second correlation data of at least one subsequent vessel at the target port; and a time prediction module for determining the expected departure time of the subsequent vessel at the target port based on the internal parameter error, the external correction error, and the second correlation data.

[0024] An embodiment of the third aspect of this application provides an electronic device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the ship departure time prediction method of the first aspect.

[0025] An embodiment of the fourth aspect of this application provides a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the ship departure time prediction method of the first aspect.

[0026] An embodiment of the fifth aspect of this application provides a chip including a processor and a communication interface, the communication interface and the processor being coupled together, the processor being used to run a program or instructions to implement the steps of the ship departure time prediction method as described in the first aspect.

[0027] Additional aspects and advantages of the technical solutions of the present invention will become apparent in the following description or may be learned by practice of the invention. Attached Figure Description

[0028] Figure 1 One of the flowcharts of the ship departure time prediction method according to this application is shown;

[0029] Figure 2 A second flowchart illustrating the ship departure time prediction method according to this application is shown.

[0030] Figure 3 The third flowchart illustrates the ship departure time prediction method according to this application;

[0031] Figure 4 The fourth flowchart illustrates the ship departure time prediction method according to this application;

[0032] Figure 5 The fifth flowchart illustrates the ship departure time prediction method according to this application;

[0033] Figure 6 A schematic block diagram of the ship departure time prediction device according to this application is shown;

[0034] Figure 7 A schematic block diagram of the structure of an electronic device according to this application is shown;

[0035] Figure 8 The sixth flowchart illustrates the method for predicting ship departure time according to this application.

[0036] Among them, 900: Ship departure time prediction device; 902: Data acquisition module; 904: Parameter determination module; 906: Model prediction module; 908: Actual parameter module; 910: Overall error module; 912: Internal error module; 914: External error module; 916: Correlation acquisition module; 918: Time prediction module;

[0037] 1000: Electronic device; 1109: Memory; 1110: Processor. Detailed Implementation

[0038] To better understand the above-described objectives, features, and advantages of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, embodiments of the invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of this application is not limited to the specific embodiments disclosed below.

[0040] The methods for predicting the estimated departure time of relevant vessels are usually based on historical statistics or the output of a single model, which have the following shortcomings:

[0041] 1. Most customers rely on the estimated time of departure (ETD) published by the shipping company. However, due to the shipping company's untimely publication of information and its neglect of factors such as port congestion and weather, customers cannot obtain accurate ETD for reasons such as self-protection.

[0042] 2. Some ETDs based on multi-data source calibration are often based on static data and cannot update and optimize the model;

[0043] 3. The prediction model cannot be continuously revised based on actual departure results;

[0044] 4. In scenarios involving multiple ports of call, errors accumulate step by step, resulting in insufficient prediction stability.

[0045] The ship departure time prediction method, device, electronic equipment, medium, and chip provided in this application will be described in detail below with reference to specific embodiments and application scenarios.

[0046] This embodiment provides a method for predicting ship departure time, such as... Figure 1 As shown, methods for predicting ship departure times include:

[0047] Step S100: Obtain the first correlation data of the target vessel at the target port;

[0048] Step S102: Determine at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first correlation data;

[0049] Step S104: Determine the estimated departure time of the target vessel based on the internal and external correction parameters;

[0050] Step S106: Obtain the actual parameters of the target vessel at the target port, including the actual departure time;

[0051] Step S108: Determine the overall prediction error based on the actual departure time and the estimated departure time;

[0052] Step S110: Determine the internal parameter error based on the actual parameters and the internal correction parameters;

[0053] Step S112: Determine the external correction error based on the overall prediction error and the internal parameter error;

[0054] Step S114: Obtain second correlation data of at least one subsequent vessel at the target port;

[0055] Step S116: Determine the estimated departure time of subsequent vessels at the target port based on internal parameter errors, external correction errors, and second correlation data.

[0056] The ship departure time prediction method provided by this invention aims to solve the problem of insufficient accuracy in predicting ship stay time in port in related technologies. Traditional methods often rely on historical statistical averages in a single dimension, failing to fully consider the variability of ship operations themselves and the dynamic impact of the complex external environment of ports, resulting in a large deviation between the prediction results and the actual situation.

[0057] This method constructs a hierarchical error correction model, which systematically decomposes the total error of departure time prediction into internal factor errors originating from the ship's own operations and external factor errors originating from the port environment. By using error analysis of ships that have completed their operations, the prediction of subsequent ships is dynamically corrected, thereby achieving high-precision and adaptive time estimation.

[0058] The target vessel refers to a vessel currently operating in the port whose departure time needs to be predicted.

[0059] The target port is the port where the target vessel is currently docked.

[0060] The first correlation data is a multi-dimensional data set whose core purpose is to comprehensively characterize the internal and external factors that affect the departure time of the target vessel.

[0061] Internal factor data typically includes detailed vessel attributes such as vessel type, tonnage, loading plan for this voyage, completed cargo loading and unloading volume, pending work list, and historical data on the vessel's own equipment status and operational efficiency.

[0062] External factor data includes real-time information from the port side, such as the port's weather conditions, tidal conditions, allocation and availability of quay cranes, tugboat and pilot scheduling plans, and traffic flow density in the port's public channels and anchorages.

[0063] For example, external factor data includes real-time and forecast wind force, wind speed, visibility and precipitation data obtained from port meteorological stations; tide tables and real-time tide height information obtained from port maritime authorities; allocation plans and current busy / idle status of key resources such as quay cranes, yard cranes, tugboats and pilots obtained from port production scheduling systems; and traffic flow density and queuing status of port channels and anchorages obtained from vessel traffic management systems.

[0064] After obtaining the first correlation data in multiple dimensions, the core quantitative parameters for the prediction model are extracted or calculated from the original data.

[0065] Specifically, at least one internal correction parameter and at least one external correction parameter of the target vessel are determined based on the first correlation data.

[0066] Internal correction parameters refer to time-related factors that are primarily determined by the ship's own operational processes.

[0067] For example, the system calculates the estimated remaining operation time based on the ship's remaining cargo handling capacity and the historical average efficiency of similar ships operating on this type of cargo. This estimated value serves as a key internal correction parameter.

[0068] External correction parameters refer to time-related effects imposed by the port environment that are beyond the control of the ship.

[0069] For example, the system assesses the potential duration of operational disruptions based on the port's gale warnings and wave height forecasts for the next few hours, combined with port operating procedures; or it calculates the waiting time for pilots based on the estimated pilot boarding time window indicated by the port's dispatch system. These assessed delay durations are the external correction parameters.

[0070] Based on the determined internal and external correction parameters, the estimated departure time of the target vessel is calculated comprehensively. The calculation logic typically uses a reference time when the vessel completes berthing or begins operations as an anchor point. It linearly superimposes the time estimates represented by the various internal correction parameters (such as operation time) with the delay estimates represented by the external correction parameters (such as waiting time and interruption time), and considers necessary buffers between operations to finally calculate a specific future time point as the estimated departure time of the target vessel.

[0071] To verify the accuracy of the prediction and drive the model's self-learning, the actual parameters of the target vessel at the target port are obtained after the vessel actually departs, among which the actual departure time is crucial.

[0072] By comparing the actual departure time with the previously predicted departure time, the overall prediction error, or total deviation, can be calculated.

[0073] Based on the overall prediction error and the internal parameter error, the external correction error can be derived in reverse. The overall prediction error can be considered to be contributed by both the internal parameter error and the external correction error. By removing the quantified internal parameter error from the overall prediction error, the remaining part can be attributed to the estimation bias of the impact of the external environment, i.e., the external correction error.

[0074] Based on the second set of correlation data, the underlying internal and external correction parameters are calculated to obtain an initial prediction value. Then, the previously calculated internal parameter error is used as a correction term to calibrate the current prediction regarding the subsequent operational efficiency of the vessel.

[0075] The system utilizes internal parameter errors and external correction errors obtained from previous analyses of vessels operating under identical or similar conditions. These errors are quantified into specific correction parameters, such as waiting time parameters to adjust for port resource waiting expectations, and port operation efficiency parameters to adjust for expected operational efficiency. Simultaneously, predictions of external environmental impacts are also revised based on historical deviations.

[0076] By using a comprehensive model, the corrected parameters are re-integrated and calculated with the second correlation data of subsequent ships, thereby outputting an optimized estimated departure time.

[0077] The estimated departure time is a predicted time point generated after multi-dimensional data fusion and dynamic error correction, which is a prediction of the departure time of a specific subsequent vessel after completing all operations at the target port.

[0078] The second associated data corresponds to the first associated data and is the sum of information extracted and spatiotemporally associated from multiple heterogeneous data sources by the subsequent ships. In other words, the second associated data is a multi-dimensional data set.

[0079] The system extracts static attributes and dynamic planning data of subsequent vessels from multiple sources, including the Automatic Identification System (AIS), port scheduling system, and vessel agent declaration. This includes inherent attributes such as vessel type, tonnage, and size, as well as information from the current voyage plan such as planned arrival time, planned workload, and reserved berths or work areas. This data forms the baseline framework for prediction.

[0080] Simultaneously, real-time and forecast data of the target port at the predicted time are accessed and correlated. This includes the port's current operational status, such as the terminal congestion index, the number of available loading and unloading equipment, and the pilotage and tugboat scheduling queues; as well as real-time meteorological and sea state data, such as wind speed, wave height, visibility, and ocean current information during the predicted operational period. This external environmental data is used to assess external conditions that may affect operational efficiency.

[0081] All data will be aligned, cleaned, and merged according to a unified time base and ship identification, forming a structured, timestamped dataset, namely the second associated data. The second associated data corresponds structurally to the first associated data used to analyze historical errors, but its content is updated with information on subsequent ships and the current environment.

[0082] Understandably, by acquiring and quantifying the internal and external factors affecting departure operations, a leap from rough estimation to accurate prediction of departure time has been achieved. The overall prediction error is decomposed into internal parameter errors originating from the ship's own operations and external correction errors originating from the port environment. This decomposition result is used to dynamically correct subsequent ship predictions, thereby constructing an intelligent prediction closed loop with self-learning and self-adaptive capabilities, which improves the accuracy and reliability of the prediction results.

[0083] This not only provides shipping companies with key decision-making support for schedule management and cost control, but also helps port operators optimize the scheduling plans of key resources such as berths, tugboats, and pilots, reducing ineffective waiting time for ships and improving overall port operation efficiency and ship turnaround speed.

[0084] In some embodiments, the internal correction parameters are optionally dynamically updated based on real-time streaming processing of vessel operation progress data in the first associated data. The system continuously receives operation volume feedback from dock cranes and shipboard loading / unloading equipment via an IoT interface, calculates instantaneous operation efficiency in real time, and uses this efficiency to continuously update the remaining operation time estimate. By transforming static parameters into dynamic parameters based on real-time data streams, it can more accurately capture acceleration or delay changes in the operation process, thereby improving the timeliness and accuracy of departure time prediction.

[0085] In some embodiments, the external correction parameter may optionally include an estimate of time delays caused by port management processes. Such delays include, but are not limited to, waiting times for procedures such as joint inspection, customs inspection of goods, and port fee payment. The system obtains queuing status and average processing time data for relevant processes by accessing the status interface of the port's single window or electronic port system, and quantifies this into an independent process delay parameter. By incorporating software and institutional factors into the external correction scope, the predictive model more comprehensively reflects various real-world constraints affecting departure times, reducing predictive blind spots caused by uncertainties in management processes.

[0086] In some embodiments, optionally, after determining the external correction error based on the overall prediction error and the internal parameter error, the system also associates the internal parameter error and the external correction error with the specific internal and external factor categories corresponding to their generation and stores them. When making predictions for subsequent ships, the system prioritizes calling historical error category data most similar to the current ship and port situation as the basis for correction. By achieving refined error attribution and contextualized matching, the error correction process is no longer a simple global offset, but a targeted, intelligent compensation based on similar scenarios, thereby significantly improving the effectiveness of correction and prediction accuracy.

[0087] In some embodiments, the determined estimated departure time is optionally provided as an input parameter to the port's resource scheduling system to optimize subsequent berth allocation, tugboat arrangement, and pilot scheduling plans. By linking the forecast output with port operation decisions in a closed loop, not only is the actual utility of the forecast results improved, but the optimized resource scheduling itself can also reduce vessel waiting time in port, thus forming a positive cycle that improves the overall port's operational efficiency and vessel turnaround speed at the system level.

[0088] Optionally, in some embodiments, when a drastic change is detected in a key external environmental data point upon which the external correction parameters depend, and this change would cause a significant inaccuracy in the estimated departure time calculated based on the original parameters, the system automatically triggers an emergency correction mode for the prediction model. In emergency correction mode, the system temporarily increases the prediction weight based on recent real-time data statistics and quickly generates a corrected estimated departure time range for user reference. By establishing a rapid response mechanism to sudden anomalies, the robustness and practicality of the prediction system are enhanced, ensuring that valuable decision support can still be provided when port operations are disrupted.

[0089] In some embodiments, optionally, such as Figure 2 As shown, step S100: Obtain the first associated data of the target vessel at the target port, including:

[0090] Step S1002: Obtain raw data from multiple heterogeneous data sources, including the target vessel's schedule data, the target port's port operation data, meteorological and sea condition data, and abnormal event data;

[0091] Step S1004: Clean the raw data to determine the preprocessed data;

[0092] Step S1006: Perform spatiotemporal alignment processing on the preprocessed data, mapping the preprocessed data corresponding to shipping schedule data, port operation data, meteorological and sea condition data and abnormal event data to a unified spatiotemporal reference system to determine the first associated data.

[0093] In this embodiment, in the face of the current situation in the shipping and port industry where data sources are scattered, formats are different, and spatiotemporal benchmarks vary, the original messy information flow is transformed into aligned, structured, and correlated data that can be directly used for quantitative analysis through preset data processing rules, laying a reliable data foundation for subsequent accurate parameter calculations and time predictions.

[0094] Raw data is obtained from multiple independent business systems, and these heterogeneous data sources together constitute the information ecosystem of the predictive model.

[0095] For example, the raw data includes, but is not limited to: the target vessel's schedule data, which usually comes from the internal management system of the shipping agency or shipping company and includes key business arrangements such as the vessel's planned arrival time, planned departure time, reserved berth, cargo volume, and cargo type.

[0096] Port operation data for the target port comes from the terminal's production operating system, equipment management system, and port dispatch center. It reflects in real time the operational status of quay cranes, the congestion of the storage yard, the dispatch queues of tugboats and pilots, and the actual occupancy and vacancy status of berths.

[0097] Meteorological and sea condition data, acquired from professional meteorological service providers, marine observation stations, and local sensor networks in ports, provides information on natural environmental factors affecting ship operations and navigation, such as wind force, wind speed, visibility, wave height, and tides.

[0098] Abnormal event data for the target port, which may come from port announcements, maritime department notices, or social media sentiment monitoring, is used to capture unplanned emergencies such as equipment failures, temporary traffic control, strikes, or severe weather warnings.

[0099] After aggregating raw data from multiple sources, data cleaning is performed to determine the usable preprocessed data.

[0100] Raw data often contains noise, missing values, outliers, and inconsistent formats. Cleaning processes include, but are not limited to: identifying and removing outliers that are clearly outside the reasonable range, such as negative ship speed records; interpolating or labeling missing timestamps or geographic coordinates based on the preceding and following data; standardizing and unifying different names for the same entity from different data sources; and converting unstructured text descriptions, such as the weather condition "sunny," into structured enumeration values.

[0101] However, cleaned data alone cannot be directly used for correlation analysis because its temporal and spatial reference systems are not yet unified.

[0102] Therefore, the system needs to perform critical spatiotemporal alignment processing on the preprocessed data.

[0103] Spatiotemporal alignment is the process of mapping all data to a unified spatiotemporal reference system.

[0104] In terms of time, the system uniformly calibrates the timestamps of all data to Coordinated Universal Time or the local port time, and aligns data with different reporting frequencies, such as Automatic Identification System (AIS) data reported every second and meteorological data updated every half hour, to a common set of analysis time points through interpolation or aggregation methods.

[0105] In terms of spatial dimensions, all data involving geographical location, such as ship coordinates, berth locations, and meteorological observation points, will be uniformly converted to the same geographical coordinate system.

[0106] More importantly, logical alignment links data from different sources according to three core dimensions: ships, ports, and time.

[0107] For example, the ship's position data at a certain moment, the status data of the quay crane that the ship is operating, and the wind speed data covering the berth area at that moment can be linked into a complete record through ship identification, berth number, and time window.

[0108] After this step is completed, the previously discrete multidimensional data is integrated into an intrinsically related, spatiotemporally synchronized data set, which is the final determined first related data that can be used for in-depth analysis.

[0109] Understandably, the data preprocessing process ensures that the information used for subsequent parameter extraction and prediction calculations is highly integrated and of high quality, fundamentally improving the stability and accuracy of the prediction model.

[0110] In some embodiments, the raw data may be cleaned, including applying cleaning rules adapted to the characteristics of the shipping schedule data, port operation data, meteorological and sea condition data, and abnormal event data respectively.

[0111] For shipping schedule data, the cleaning rules focus on identifying and correcting logical inconsistencies, such as departure time being earlier than arrival time.

[0112] For port equipment status data, the focus is on filtering out jitter signals caused by momentary sensor malfunctions.

[0113] For meteorological and marine data, cleaning rules may include imputing missing local station data based on geospatial correlation.

[0114] By implementing differentiated, domain-knowledge-based cleaning strategies, it is possible to specifically remove major noise from various data sources and improve the overall credibility of preprocessed data.

[0115] In some embodiments, optional spatiotemporal alignment processing is implemented, the core of which is mapping all preprocessed data to a unified spatiotemporal framework. The unified spatiotemporal framework uses Coordinated Universal Time (UTC) as the base time axis and the latitude and longitude of the berth where the target vessel is scheduled to call as the spatial reference origin. Data such as port equipment status and meteorological observations are dynamically determined based on their spatial distance from the reference origin and the representativeness of the data itself, to determine whether they should be associated with the vessel's status at the current time or a set time, thereby achieving precise data linkage between the vessel, the port, and the environment.

[0116] By establishing a unified reference system centered on the prediction target, it is ensured that subsequent analyses use effective information that is strongly correlated with the target ship in time and space, while discarding irrelevant environmental noise.

[0117] In some embodiments, optionally, after determining the first associated data, the method further includes performing consistency verification and conflict resolution on the data set corresponding to the first associated data.

[0118] For example, when there is a significant discrepancy between the planned departure time from the shipping company and the estimated operation completion time from the terminal operating system, the system will assign different weights to the real-time performance, authority, and historical accuracy of the data sources, and calculate a reconciliation value using a confidence-weighted approach. Alternatively, it may use information from the more real-time data source to cover the less real-time data. Through a built-in consistency arbitration mechanism, common contradictions between multi-source data can be automatically resolved, generating a set of logically consistent correlated data to provide stable and reliable input for the predictive model.

[0119] In some embodiments, optionally, the process of acquiring, cleaning, and aligning is continuously performed while the target vessel is in port, so that the first associated data can dynamically reflect the progress of operations, resource changes, and environmental changes.

[0120] In some embodiments, optionally, such as Figure 3 As shown, step S102: Determine at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first correlation data, including:

[0121] Step S1020: Determine the port waiting time based on the preprocessed data corresponding to the port operation data in the first associated data;

[0122] Step S1022: Determine the predicted operation time corresponding to the target vessel based on the preprocessed data corresponding to the ship schedule data and port operation data in the first associated data;

[0123] Step S1024: Determine the internal correction parameters based on the port waiting time and the predicted operation time.

[0124] In this embodiment, the cleaned and aligned high-quality first correlation data is transformed into internal correction parameters for quantifying the time spent by vessels on core operations in port. These internal correction parameters directly reflect the time investment required for the target vessel to complete its planned operations.

[0125] Internal correction parameters are a quantitative expression of the total time a vessel spends in port due to its own operational processes. The time a vessel spends from being ready to operate to completing all operations and preparing to depart mainly consists of two consecutive stages driven by different factors.

[0126] The first stage is the passive waiting experienced by ships in order to obtain and occupy the necessary port resources and services. Its duration depends on the resource scheduling and allocation status of the port system, from which port waiting time is derived.

[0127] The second stage is the active operation of the ship after it has occupied resources and carried out productive operations such as loading, unloading and replenishment. Its duration depends on the scale of the task and the efficiency of the operation, from which the predicted operation duration is derived.

[0128] Determining port waiting time involves a process of derivation from port operation data to time series status, and then to cumulative waiting time.

[0129] Its input is port operation data that has been aligned in time and space. Port operation data essentially reflects the planned occupancy status and actual busy status of various port resources on the time axis.

[0130] First, the system identifies the sequence of critical port services and resources that the target vessel must rely on before departure, such as pilotage, tugboats, berths, and joint inspections. Then, for each item in the sequence, the system extracts two key states from the preprocessed data: the earliest available time when the resource can provide services to the target vessel, and the time when the target vessel requests or plans to use the resource. The difference between these two values ​​represents the waiting time for a particular resource.

[0131] Because there are sequential or parallel relationships between various services, the system, based on a pre-defined port operation process model, logically arranges and de-overlaps these discrete waiting time periods for individual resources, ultimately synthesizing a coherent total port waiting time. This time represents the time delay imposed on ship progress by external resource availability, i.e., the port waiting time.

[0132] Port waiting time is not a single value, but a combination of multiple sub-waiting times. Its determination process is a calculation based on real-time status and scheduling rules.

[0133] For example, the queuing sequence and estimated waiting time of the target vessel waiting for the pilot to guide it into the port are obtained from the port vessel traffic management system; the estimated arrival time of the tugboat serving the vessel is obtained from the tugboat scheduling system, thereby calculating the time window for waiting for the tugboat; if the vessel needs to wait for the berth to become available, the difference between its planned berthing time and the current time is obtained from the berth allocation plan.

[0134] The inputs for predicting operation duration mainly depend on the scale of the operation tasks provided by the shipping schedule data, as well as the operational capacity configuration reflected in the port operation data.

[0135] Specifically, the amount of physical work that needs to be done is extracted from the shipping schedule data, such as the number of containers or the tonnage of cargo.

[0136] At the same time, the operational resource units allocated to ships and their efficiency parameters, such as the number of quay cranes and the average operating efficiency of a single crane, are identified from the port operation data.

[0137] The work efficiency parameter may be a historical statistical value or a dynamic value adjusted according to real-time working conditions.

[0138] Divide the total workload by the product of the number of resource units and the efficiency parameter to obtain the theoretical duration of the core task.

[0139] This calculation model transforms static planned task quantities into dynamic predicted task durations, which directly quantify the time investment necessary to complete core production activities.

[0140] Predicted operation time refers to the total time that a vessel is expected to spend on productive operations such as loading and unloading cargo, refueling, and maintenance at its berth. It directly reflects the intensity and time required for a vessel to complete its core business mission.

[0141] Ultimately, the system determines internal correction parameters based on the calculated port waiting time and predicted operation time. The determination of these internal correction parameters includes, but is not limited to, directly adding the port waiting time and predicted operation time, as both together cover the main time consumption between a vessel completing arrival formalities and being ready to depart.

[0142] In more complex models, the internal correction parameters may be presented as a baseline timeline, which marks the expected start and end times of each waiting and operation phase.

[0143] Regardless of its composition, the essence of the internal correction parameters is to quantify and structure the time spent on key activities of ships within port, providing a solid foundation for generating a total forecast by superimposing it with external factors. Through this decomposition and calculation, the internal correction parameters are no longer fuzzy empirical estimates, but rather derivations based on real-time data and explicit rules, significantly improving the objectivity and accuracy of the forecast.

[0144] In some embodiments, the port waiting time is optionally determined by performing time-series conflict detection and analysis on port operation data. The system analyzes the planned timeline of the target vessel and the timeline of port resource occupancy, identifies all time intervals in which the target vessel must wait due to resources being occupied by preceding tasks, and concatenates and merges these intervals according to the logical order of the work process to obtain the total waiting time. By basing the calculation of the waiting time on explicit resource timeline conflict detection, rather than simple empirical estimation, the determination of parameters becomes more objective and interpretable.

[0145] In some embodiments, the port waiting time and predicted operation time are optionally recalculated periodically based on the latest port operation data and operation progress data, and the internal correction parameters are updated on a rolling basis. By transforming the internal correction parameters from static values ​​to a dynamic sequence that evolves over time, the prediction of departure time can reflect the latest progress at the operation site in real time, achieving continuity and adaptability in the prediction.

[0146] In some embodiments, step S102, which involves determining at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first associated data, further includes determining the external correction parameter based on the preprocessed data corresponding to the meteorological and sea state data and the abnormal event data in the first associated data.

[0147] In this embodiment, it is clarified that the external correction parameters are mainly determined based on two types of information reflecting the port's external environment and unforeseen circumstances: meteorological and sea state data and abnormal event data. These external correction parameters are used to quantify external factors that are not directly controlled by the ship and port's operational plans but can interfere with their processes and cause additional time delays.

[0148] External and internal correction parameters are logically complementary, together forming a complete prediction of the total time a ship spends in port. Internal correction parameters are based on planned tasks and resource scheduling, while external correction parameters respond to unplanned environmental disturbances and unforeseen events.

[0149] The external correction parameters rely on the meteorological and marine data in the first associated data, which is a set of data that has been spatiotemporally aligned and reflects the physical environment and forecast information of the target port and surrounding sea areas in the future, such as time series data of wind, visibility, waves and tides.

[0150] Abnormal event data refers to information that indicates sudden, unplanned events that disrupt the normal operation of the port, such as notices of sudden equipment failures, safety accidents, temporary traffic control, and labor actions.

[0151] Converting meteorological and sea condition data into time delay parameters involves a matching and calculation process based on a pre-set base of port operation safety and efficiency rules.

[0152] The rule base defines the feasibility of different types of port operations, the efficiency reduction factor, or the threshold for necessary interruption under different levels of weather and sea conditions.

[0153] The system compares time-series meteorological forecast data with these rules on a time-by-time basis. When the forecast data reaches or exceeds the interruption threshold of a certain operational stage, the time period is directly marked as a forced interruption delay.

[0154] When the forecast data falls within a range that may lead to reduced efficiency, the system invokes the corresponding efficiency reduction model and calculates the equivalent delay time based on the reduction factor.

[0155] In addition, for some ports or large vessels that are restricted by tidal windows, it is also necessary to calculate the available operating time window based on tidal forecast data.

[0156] The abnormal events are analyzed in a structured manner to extract key attributes such as event type, scope of impact, severity level, and estimated handling time.

[0157] Then, the event type is indexed to a pre-defined event impact mapping model.

[0158] The event impact mapping model defines which subsystems or operational processes in port operations are typically affected by such events, and the mode of impact, whether it leads to complete disruption, reduced efficiency, or process reordering.

[0159] Based on the attributes of the event, the affected operational processes and their expected delays can be calculated.

[0160] For example, the impact model of a "quay crane main component failure" event would point to the interruption of loading and unloading operations at a specific berth, and the delay time could be estimated based on historical data of maintenance resource allocation and the complexity of the failure.

[0161] The multiple meteorological and event-related impact components calculated above are combined into a complete external correction parameter. This combination is not a simple addition but requires temporal and logical integration. By analyzing the predicted occurrence time windows of each impact component, their overlap with the internal operational timeline is determined, and concurrent or coupled effects between multiple external factors are addressed to avoid redundant calculations.

[0162] External correction parameters can be expressed as an external impact timeline, which marks the start and end times, cause types, and impact degrees of various delay intervals caused by external factors during the target vessel's expected port stay.

[0163] In some embodiments, the meteorological and marine state data may optionally include fused information obtained from multiple data sources with different spatiotemporal resolutions and forecast durations. The system employs data assimilation technology to fuse heterogeneous data from meteorological stations, satellite remote sensing, and numerical weather prediction models to generate a spatiotemporally continuous and more confident comprehensive meteorological field for the target port area.

[0164] Subsequent impact assessments based on this fused, high-precision meteorological field can improve the accuracy of capturing complex weather phenomena, thereby making the external correction parameters determined accordingly more reliable.

[0165] In some embodiments, optionally, such as Figure 4 As shown, step S110: The actual parameters also include the actual waiting time and the actual operation time. The internal parameter error is determined based on the actual parameters and the internal correction parameters, including:

[0166] Step S1102: Determine the waiting time error based on the actual waiting time and the port waiting time;

[0167] Step S1104: Determine the operation time error based on the actual operation time and the predicted operation time.

[0168] In this embodiment, after the target vessel actually departs, the internal correction parameters on which the previous prediction was based are accurately compared with the final actual operation time data, thereby quantifying the deviation of the prediction model in assessing the time consumption of the vessel's own operation process.

[0169] The determination of the intrinsic parameter error is based on a difference comparison of two sets of time data. The first set of data consists of internally corrected parameters generated during the forecasting phase, which includes the predicted port waiting time and predicted operation time. The second set of data consists of actual parameters obtained after the actual operation is completed, which supplements the actual waiting time and actual operation time. The intrinsic parameter error is determined by calculating the difference between the corresponding components in these two sets of data.

[0170] Determining the actual waiting time requires continuously tracking and recording all idle periods that the target vessel experiences while waiting for port resources, from the moment it arrives at the port control area until it actually begins to carry out core productive operations.

[0171] This is achieved by comparing timestamp sequences, such as recording the time when the ship actually receives pilot service, the time when it actually berths, and the time when it actually begins loading and unloading operations. The difference between the start time of each service and the time when the prerequisite conditions are met is the actual waiting time for each item, and their sum constitutes the actual waiting time.

[0172] Port waiting time is the sum of similar waiting times estimated during the forecasting phase based on resource scheduling plans and status. Waiting time error is calculated by subtracting the arithmetic difference between the actual waiting time and the total port waiting time.

[0173] The determination of actual operation time requires recording the net time consumed from the start of the first loading, unloading or resupply operation to the completion of all scheduled operations and the vessel being ready to depart.

[0174] This is typically calculated by integrating data from the port's operating system, obtaining the actual start and end timestamps of each work instruction, and deducting any planned interruptions due to rest, meals, or external factors.

[0175] Predicted job duration is the estimated job time calculated during the forecasting phase based on job volume, resource allocation, and efficiency models.

[0176] The task duration error is obtained by calculating the arithmetic difference between the actual task duration and the predicted task duration.

[0177] Error values ​​quantify the deviations in assessing the efficiency of a ship's specific operational tasks, reflecting the accuracy of estimates of operational rate, equipment performance, or task complexity.

[0178] By calculating the waiting time error and the operation time error separately, a refined decomposition of the internal parameter errors was completed. This clarified whether the prediction deviation mainly stemmed from misjudgments of the port system's response speed or misjudgments of the vessel's on-site operational capabilities, thus enabling more targeted subsequent corrective measures.

[0179] In some embodiments, the actual waiting time and the actual working time are not a single total duration value, but are recorded as a detailed sequence of timestamps consisting of multiple sub-periods.

[0180] The actual waiting time sequence records the specific start and end times of waiting for various resources such as pilotage, tugboats, berths, and joint inspections.

[0181] The actual operation time sequence records the specific start and end times of various operations such as loading, unloading, resupply, and maintenance.

[0182] By adopting a serialized recording method, subsequent waiting time errors and operation time errors can be located to more specific operation stages, thereby providing a spatiotemporal data foundation for refined error attribution and model calibration.

[0183] In some embodiments, optionally, the waiting time error and the job duration error, after being calculated, will be further associated with and stored in relation to the specific time period in which the error occurred, the resource type involved, or the job type.

[0184] By contextualizing and classifying errors in multiple dimensions, the most relevant historical error data can be used for targeted correction when predicting new ships in similar situations, thereby improving the accuracy and effectiveness of the error feedback mechanism.

[0185] In some embodiments, the actual waiting time is optionally determined by fusing multi-source timestamp data from port IoT sensors, ship agent reports, and terminal operating system logs, followed by conflict verification and evidence fusion. By comparing the timestamps of the same event recorded from different sources, a weighted calculation is performed based on the historical reliability of the data sources to arrive at a credible actual event occurrence time, thereby calculating a high-confidence actual waiting time.

[0186] In some embodiments, optionally, such as Figure 5 As shown, step S116: Determine the estimated departure time of subsequent vessels at the target port based on internal parameter errors, external correction errors, and second correlation data, including:

[0187] Step S1160: Obtain the preset weight coefficients;

[0188] Step S1162: Determine the correction parameters based on the weighting coefficients, external correction parameters, and external correction errors;

[0189] Step S1164: Determine the port operation efficiency parameters based on the weighting coefficient, port waiting time, and actual waiting time;

[0190] Step S1166: Determine the waiting time parameter based on the weighting coefficient, predicted operation time, and actual operation time;

[0191] Step S1168: Determine the estimated departure time of subsequent vessels based on the correction parameters, port operation efficiency parameters, waiting time parameters, and the second correlation data.

[0192] In this embodiment, the prediction error knowledge obtained from the analysis of the preceding target ship is applied to the prediction of subsequent ships through a set of quantitative correction models, thereby achieving iterative improvement in prediction accuracy.

[0193] A predictive correction model with multiple correction sub-items is established. The predictive correction model uses the newly acquired second correlation data of subsequent vessels as the basic input and various parameters obtained from historical error analysis as correction factors. Through weighted fusion, it generates an estimated departure time corresponding to the subsequent vessels.

[0194] The weighting coefficients are pre-set adjustment parameters used to balance the weight of historical experience and current real-time information in the prediction.

[0195] The preset weighting coefficients may include the confidence weight of historical error data, the sensitivity weight of different types of errors, etc.

[0196] A comprehensive correction parameter is determined based on the obtained weighting coefficients, the external correction parameters calculated for the preceding vessels, and the external correction error.

[0197] The external correction error reflects the previous estimation bias of the impact of the external environment, and the system also generates an external correction parameter based on its real-time data for subsequent ships.

[0198] Determining the correction parameters involves converting the previous external correction error into a correction value according to a certain weighting coefficient, and then applying it to the new external correction parameters.

[0199] This process can be understood as using historical error experience to correct current judgments about similar external factors, thereby compensating for possible systemic biases in the system.

[0200] For example, if history shows that the delayed effects of a certain weather event are generally underestimated, then the estimated impact of similar weather events will be adjusted upwards by a certain percentage when forecasting subsequent ships.

[0201] Similarly, the system uses weighting coefficients and combines them with port waiting time, actual waiting time, predicted operation time, and actual operation time obtained from the analysis of preceding vessels to determine two key efficiency parameters: port operation efficiency parameter and waiting time parameter.

[0202] Port operation efficiency parameters quantify the resource scheduling and supply efficiency of the port system in actual operation.

[0203] The calculation of port operation efficiency parameters compares the predicted operation time of preceding vessels with the actual operation time.

[0204] If the actual operation time is consistently shorter than the forecast, it may indicate that the port's current operational efficiency is higher than the model baseline; conversely, it may indicate lower efficiency.

[0205] Based on the weighting coefficients, this historical efficiency deviation information is integrated to generate a port operation efficiency parameter.

[0206] Waiting time parameter quantifies the regular deviation between the actual waiting time experienced by ships and the predicted waiting time in order to obtain various services at a specific port and during a specific time period.

[0207] The waiting time parameter is based on the port waiting time of previous vessels and the actual waiting time. By analyzing historical waiting deviation patterns and considering weighting coefficients, the system generates a waiting time parameter.

[0208] The waiting time parameter reflects the waiting time offset caused by relatively stable factors such as port congestion and scheduling strategies. This parameter is used to correct for the baseline port waiting time estimate when forecasting subsequent vessels.

[0209] The correction parameters, port operation efficiency parameters, waiting time parameters, and the second correlation data of subsequent vessels are input into the prediction model to determine their final estimated departure time.

[0210] Based on the second correlation data, using the same baseline model as the predicted preceding vessels, a preliminary, uncorrected estimated departure time was calculated.

[0211] Then, the port operation efficiency parameter is applied to correct its operation efficiency assumption, the waiting time parameter is applied to correct its resource waiting assumption, and the correction parameter is applied to correct its external risk assumption.

[0212] These correction operations essentially utilize historical error feedback to dynamically adjust the various input parameters or intermediate variables of the baseline model.

[0213] Through this multi-faceted parametric correction, the new prediction results are no longer isolated, but inherit and apply the knowledge verified in historical operations, thereby significantly improving their adaptability and accuracy.

[0214] In some embodiments, the preset weighting coefficients are optionally not fixed values, but are dynamically calculated based on the historical statistical characteristics of external correction errors and internal parameter errors.

[0215] The system analyzes the mean, variance, and trends of various errors in recent ship forecasts to dynamically adjust the level of confidence in historical error data and update the weighting coefficients.

[0216] By adapting the weighting coefficients to the recent performance of the prediction error, the model can automatically balance its tendency to follow historical patterns and respond to the latest changes, thereby improving the adaptability and robustness of the correction strategy.

[0217] In some embodiments, the determination of port operation efficiency parameters and waiting time parameters may depend not only on data from a single preceding vessel, but also on joint analysis of error sequences of multiple vessels of the same type or at the same berth within a sliding time window.

[0218] The average deviation and confidence interval between operational efficiency and waiting time within a window are used as the basis for determining the current port operational efficiency and waiting time parameters. Through statistical analysis based on vessel queues, the random fluctuations in individual vessel operations can be effectively smoothed out, and more stable and reliable efficiency parameters reflecting the current overall operational status of the port can be extracted.

[0219] In some embodiments, the final output of the expected departure time may be a probability distribution or time interval, rather than a single point in time.

[0220] The width of the probability distribution or interval is calculated by comprehensively using an error propagation model based on the uncertainties of the correction parameters, port operation efficiency parameters, waiting time parameters, and the uncertainty of the second associated data itself.

[0221] By providing forecasts with uncertainty metrics, it offers richer and more risk-management-value decision-making information for vessel scheduling and port operation planning.

[0222] In one specific embodiment, the present invention optionally adopts a four-layer architecture design to construct a complete technical process from multi-source data acquisition, data processing, predictive modeling to dynamic calibration based on actual arrival feedback. Through multi-source data collaboration and parameter dynamic adjustment mechanism, the accuracy and reliability of departure time (ETD) prediction at the port of origin are improved.

[0223] The four-layer architecture consists of the following layers: data layer, processing layer, model layer, and calibration layer.

[0224] Data Layer: Multi-source heterogeneous data acquisition and access

[0225] The data layer is used to collect and integrate multi-dimensional data affecting vessel departure times, serving as the basic input for ETD prediction and calibration. This data includes, but is not limited to:

[0226] 1. Shipping company data:

[0227] This includes bill of lading-level cargo tracking data, dynamic ETD issued by shipping companies, and their historical change records;

[0228] 2. Shipping schedule data:

[0229] This includes liner schedule information, vessel name and voyage number, port call sequence, and planned departure time;

[0230] 3. Port operation data:

[0231] This includes statistics on port berth utilization, loading and unloading operation status, historical operation efficiency, and waiting time.

[0232] 4. Meteorological and sea state data:

[0233] This includes weather forecasts for the future, especially extreme weather events such as typhoons, strong winds and waves, and tsunamis that may affect navigation safety and speed.

[0234] 5. Port anomaly event data:

[0235] This includes port strikes, congestion, temporary policy changes, and other unforeseen events that may cause ship delays;

[0236] 6. Other auxiliary data:

[0237] This includes factors such as ship type, ship size, route attributes, and seasonality.

[0238] The above data is accessed into the system through a unified interface, forming the data foundation pool for ETD prediction and calibration.

[0239] Processing layer: Data quality control and spatiotemporal alignment:

[0240] The processing layer is used to resolve inconsistencies in format, accuracy, and temporal granularity among multi-source data to ensure data reliability in subsequent prediction and calibration processes.

[0241] The processing layer includes the following processing steps:

[0242] 1. Data cleaning and processing:

[0243] Identify and remove abnormal data, and complete missing data based on historical statistical characteristics or similar ship data;

[0244] 2. Data standardization processing:

[0245] Unify time expression formats (such as timestamps), geographic coordinate systems, numerical units, and port coding rules;

[0246] 3. Spatiotemporal alignment processing:

[0247] Data from different sources and with different update frequencies are mapped to a unified spatiotemporal reference system, such as using the ship's current position as the reference spatial point and the predicted time as the time axis, to form a fusionable multidimensional feature set.

[0248] The above processing ensures that the data entering the prediction model is consistent and comparable.

[0249] Model layer: Multi-source fusion prediction algorithm:

[0250] The model layer is used to predict and calculate the various component parameters of the originating port ETD.

[0251] In this invention, the originating ETD is broken down into multiple independently predictable components, including:

[0252] 1. Port waiting time parameters:

[0253] Based on historical port waiting data and recent operational status, predict the time a vessel will spend at port anchorage or waiting for berth.

[0254] 2. Port operation time parameters:

[0255] Based on the port's historical operational efficiency, vessel type and size, predict the loading and unloading operation time after berthing;

[0256] 3. External correction parameters:

[0257] This includes ETD deviation correction parameters for shipping companies, weather impact correction parameters, and abnormal event impact parameters.

[0258] The above parameters are calculated using a multi-source feature fusion model. The model can be implemented using statistical learning or machine learning methods and supports a rolling update mechanism based on the latest data, thereby outputting phased ETD prediction results.

[0259] Calibration layer: ATD and ATB are dynamically adjusted to participate in the calculation of ETD parameters.

[0260] The calibration layer is a key technical module of this invention, used to dynamically adjust the various component parameters of ETD output by the model layer after obtaining the actual time of departure (ATD) and actual time of berthing (ATB).

[0261] The estimated departure time is expressed as:

[0262] ETD_pred =T_wait_pred +T_work_pred +T_ext_pred;

[0263] Where T_wait_pred is the predicted waiting time, T_work_pred is the predicted job duration, and T_ext_pred is the external correction parameter.

[0264] Specifically, the following steps are included:

[0265] 1. Obtaining actual results:

[0266] Once the vessel or cargo actually arrives at the port of origin, the system obtains the corresponding actual departure time (ATD), actual berthing time (ATB), actual waiting time, actual operation time, and other execution results.

[0267] T_wait_real =ATB_real-ATA_real;

[0268] Where T_wait_real is the actual waiting time of the vessel, ATB_real is the actual berthing time, and ATA_real is the actual arrival time.

[0269] T_work_real =ATD_real-ATB_real;

[0270] Where T_work_real is the actual working time and ATD_real is the actual departure time.

[0271] 2. Prediction error decomposition:

[0272] The overall error between the predicted ETD and the actual ATD is decomposed into corresponding parts such as port waiting time, port operation time and external correction parameters;

[0273] Overall error calculation:

[0274] E_total = ATD_real - ETD_pred;

[0275] Where E_total is the total error, ETD_pred is the estimated departure time, and ATD_real is the actual departure time.

[0276] Error decomposition:

[0277] Waiting time error:

[0278] E_wait =T_wait_real-T_wait_pred;

[0279] Where E_wait is the waiting time error, T_wait_real is the actual ship waiting time, and T_wait_pred is the predicted waiting time.

[0280] Error in operation time:

[0281] E_work =T_work_real-T_work_pred;

[0282] Where E_work is the job duration error, T_work_real is the actual job duration, and T_work_pred is the predicted job duration.

[0283] External correction error:

[0284] E_ext =E_total-E_wait-E_work;

[0285] Where E_ext is the external correction error.

[0286] 3. Parameters are dynamically updated:

[0287] Based on the above error decomposition results, the corresponding parameter model is updated, including:

[0288] Waiting time parameter update; T_wait_new = (1-α)T_wait_pred+α×T_wait_real;

[0289] Port operation efficiency parameters update; T_work_new = (1-β)T_work_pred+β×T_work_real;

[0290] Parameter correction is affected by weather or abnormal events; T_ext_new = (1-γ)T_ext_pred+γ×E_ext.

[0291] Wherein, α, β, and γ are preset weighting coefficients, and α+β+γ=1.

[0292] 4. Closed-loop feedback mechanism:

[0293] The updated parameters will directly participate in subsequent ETD prediction calculations, enabling the prediction model to continuously adapt to changes in the real operating environment over time, forming a closed-loop self-learning mechanism based on actual results.

[0294] By designing a calibration layer, this invention avoids the long-term accumulation problem of ETD prediction errors and significantly improves the stability and accuracy of prediction results under different routes, ports and time windows.

[0295] It is understandable that ETD can be viewed as learnable, capable of dynamically adjusting parameters based on the actual departure time (ATD) and actual berthing time (ATB), thus providing feedback correction to ETD.

[0296] In one specific embodiment, such as Figure 8 As shown, methods for predicting ship departure times also include:

[0297] Step S200: Obtain the port of origin, vessel name, and maritime mobility service identification code for which ETD is to be calculated;

[0298] Step S202: Obtain multi-source data and organize the calculation of ETD parameters;

[0299] Step S204: Obtain the vessel name and sailing date ETD;

[0300] Step S206: Obtain Cargo Tracking ETD;

[0301] Step S208: Obtain the operation duration at the port of origin;

[0302] Step S210: Obtain the waiting time at the port of origin anchorage;

[0303] Step S212: Obtain future weather conditions;

[0304] Step S214: Obtain various abnormal events such as port strikes;

[0305] Step S216: Dynamically adjust parameters based on actual ATD and actual ATB;

[0306] Step S218: Calculate the ETD by weighting each parameter.

[0307] The method for predicting ship departure time begins with obtaining the basic identification information required for the calculation, namely the target ship's name, maritime mobility service identification code, and its port of origin.

[0308] After obtaining the basic information, the process enters the core stage of multi-source data acquisition and organization. This involves aggregating various parameters required for ETD calculation from multiple independent but related business systems.

[0309] Specifically, these parameters include, but are not limited to: vessel name and schedule ETD obtained from shipping company or agent plans; cargo tracking ETD obtained from port or terminal operating systems; departure anchorage waiting time obtained from port anchorage queuing and scheduling rules analysis; departure operation time estimated based on historical data of vessel workload, quay crane allocation and efficiency; future weather forecast data from meteorological service agencies; and information on sudden abnormal events that may affect port operations, such as strikes and equipment failures.

[0310] Different confidence weights or priorities are assigned to the forecast times from different sources, such as shipping ETD and cargo tracking ETD, as well as time components such as operation time and waiting time. A comprehensive ETD estimate is then generated through a weighted fusion model.

[0311] Actual departure time (ATD) and actual berthing time (ATB) are the actual data reported after the vessel has completed its operations.

[0312] By comparing historical forecasts with actual values, the model coefficients used to calculate internal parameters such as operation time and waiting time are dynamically adjusted, or the evaluation weights of external factors (such as weather effects) are corrected, so that the prediction model can learn and optimize itself and continuously improve the accuracy of future ship ETD calculations.

[0313] like Figure 6 As shown in the illustration, this application embodiment also provides a ship departure time prediction device 900, which includes: a data acquisition module 902 for acquiring first correlation data of a target ship at a target port; a parameter determination module 904 for determining at least one internal correction parameter and at least one external correction parameter of the target ship based on the first correlation data; a model prediction module 906 for determining the expected departure time of the target ship based on the internal correction parameter and the external correction parameter; an actual parameter module 908 for acquiring actual parameters of the target ship at the target port, including the actual departure time; an overall error module 910 for determining an overall prediction error based on the actual departure time and the expected departure time; an internal error module 912 for determining an internal parameter error based on the actual parameters and the internal correction parameter; an external error module 914 for determining an external correction error based on the overall prediction error and the internal parameter error; a correlation acquisition module 916 for acquiring second correlation data of at least one subsequent ship at the target port; and a time prediction module 918 for determining the expected departure time of the subsequent ship at the target port based on the internal parameter error, the external correction error, and the second correlation data.

[0314] like Figure 7As shown, this application embodiment also provides an electronic device 1000, including a processor 1110, a memory 1109, and a program or instructions stored in the memory 1109 and executable on the processor 1110. When the program or instructions are executed by the processor 1110, they implement the various processes of the above-described embodiment of the ship departure time prediction method and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0315] Optionally, the processor 1110 is used to acquire the first associated data of the target vessel at the target port;

[0316] Optionally, the processor 1110 is further configured to determine at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first correlation data;

[0317] Optionally, the processor 1110 is also configured to determine the estimated departure time of the target vessel based on internal correction parameters and external correction parameters;

[0318] Optionally, the processor 1110 is also used to obtain the actual parameters of the target vessel at the target port, including the actual departure time.

[0319] Optionally, the processor 1110 is also configured to determine the overall prediction error based on the actual departure time and the expected departure time;

[0320] Optionally, the processor 1110 is also configured to determine the internal parameter error based on the actual parameters and the internal correction parameters;

[0321] Optionally, the processor 1110 is also configured to determine the external correction error based on the overall prediction error and the internal parameter error;

[0322] Optionally, the processor 1110 is also configured to acquire second correlation data of at least one subsequent vessel at the target port;

[0323] Optionally, the processor 1110 is also configured to determine the estimated departure time of subsequent vessels at the target port based on internal parameter errors, external correction errors, and second correlation data.

[0324] The memory 1109 can be used to store software programs and various data. The memory 1109 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 1109 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 1109 in this embodiment includes, but is not limited to, these and any other suitable types of memory.

[0325] This application also provides a readable storage medium storing a program or instructions. When executed by a processor, the program or instructions implement the various processes of the above-described ship departure time prediction method embodiments and achieve the same technical effects. To avoid repetition, these will not be described again here. Furthermore, the readable storage medium improves the data storage capacity and data processing speed of the ship departure time prediction method in this application.

[0326] A computer-readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium can be an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing, but is not limited thereto. A non-exhaustive list of more specific examples of computer-readable storage media includes: portable computer floppy disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable optical disc read-only memory (CD-ROM), digital universal disk (DVD), memory cards, floppy disks, encoding mechanical devices (e.g., punched cards or grooves with raised structures for recording instructions), and any suitable combination of the foregoing. The computer-readable storage medium used herein should not be construed as the transmission of signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media, or electrical signals transmitted through wires.

[0327] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0328] This application also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described ship departure time prediction method embodiments, and can achieve the same technical effect. To avoid repetition, it will not be described again here. In addition, the chip improves the data processing speed of the ship departure time prediction method in this application.

[0329] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0330] In this invention, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance; the term "multiple" refers to two or more unless otherwise explicitly defined. The terms "install," "connect," "link," and "fix" should be interpreted broadly. For example, "connect" can be a fixed connection, a detachable connection, or an integral connection; "link" can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0331] In the description of this invention, it should be understood that the terms "upper," "lower," "left," "right," "front," "rear," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or unit referred to must have a specific orientation or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0332] In the description of this specification, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to specific features, structures, materials, or characteristics described in connection with an embodiment or example that are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0333] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting ship departure time, characterized in that, include: Obtain the first relevant data of the target vessel at the target port; Based on the first correlation data, at least one internal correction parameter and at least one external correction parameter of the target vessel are determined; The estimated departure time of the target vessel is determined based on the internal correction parameters and the external correction parameters; Obtain the actual parameters of the target vessel at the target port, including the actual departure time; The overall prediction error is determined based on the actual departure time and the estimated departure time. The internal parameter error is determined based on the actual parameters and the internal correction parameters; The external correction error is determined based on the overall prediction error and the internal parameter error; Obtain second correlation data for at least one subsequent vessel at the target port; The estimated departure time of the subsequent vessel at the target port is determined based on the internal parameter error, the external correction error, and the second correlation data.

2. The method for predicting ship departure time according to claim 1, characterized in that, The acquisition of the first associated data of the target vessel at the target port includes: Acquire raw data from multiple heterogeneous data sources, including the target vessel's schedule data, the target port's port operation data, meteorological and sea condition data, and abnormal event data; The raw data is cleaned to determine the preprocessed data; The preprocessed data is spatiotemporally aligned to map the preprocessed data corresponding to the shipping schedule data, the port operation data, the meteorological and sea condition data, and the abnormal event data to a unified spatiotemporal reference system to determine the first associated data.

3. The method for predicting ship departure time according to claim 2, characterized in that, The step of determining at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first associated data includes: The port waiting time is determined based on the preprocessed data corresponding to the port operation data in the first associated data. The predicted operation time corresponding to the target vessel is determined based on the preprocessed data corresponding to the shipping schedule data and port operation data in the first associated data. The internal correction parameters are determined based on the port waiting time and the predicted operation time.

4. The method for predicting ship departure time according to claim 2, characterized in that, The step of determining at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first associated data includes: The external correction parameters are determined based on the preprocessed data corresponding to the meteorological and sea condition data and the abnormal event data in the first associated data.

5. The method for predicting ship departure time according to claim 3, characterized in that, The actual parameters also include the actual waiting time and the actual operation time. Determining the internal parameter error based on the actual parameters and the internal correction parameters includes: The waiting time error is determined based on the actual waiting time and the port waiting time. The operation time error is determined based on the actual operation time and the predicted operation time.

6. The method for predicting ship departure time according to any one of claims 1 to 5, characterized in that, Determining the estimated departure time of the subsequent vessel at the target port based on the internal parameter error, the external correction error, and the second correlation data includes: Obtain the preset weight coefficients; The correction parameters are determined based on the weighting coefficients, the external correction parameters, and the external correction error. The port operation efficiency parameters are determined based on the weighting coefficients, port waiting time, and actual waiting time. The waiting time parameter is determined based on the weighting coefficient, the predicted operation time, and the actual operation time. The estimated departure time of the subsequent vessel is determined based on the correction parameters, the port operation efficiency parameters, the waiting time parameters, and the second associated data.

7. A device for predicting ship departure time, characterized in that, include: The data acquisition module is used to acquire the first relevant data of the target vessel at the target port; The parameter determination module is used to determine at least one internal correction parameter and at least one external correction parameter of the target vessel based on the first associated data. The model prediction module is used to determine the expected departure time of the target vessel based on the internal correction parameters and the external correction parameters. The actual parameter module is used to obtain the actual parameters of the target vessel at the target port, including the actual departure time. The overall error module is used to determine the overall prediction error based on the actual departure time and the expected departure time; An internal error module is used to determine the internal parameter error based on the actual parameters and the internal correction parameters; An external error module is used to determine an external correction error based on the overall prediction error and the internal parameter error; The association acquisition module is used to acquire second association data of at least one subsequent vessel at the target port; The time prediction module is used to determine the estimated departure time of the subsequent vessel at the target port based on the internal parameter error, the external correction error, and the second correlation data.

8. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the ship departure time prediction method as described in any one of claims 1 to 6.

9. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the ship departure time prediction method as described in any one of claims 1 to 6.

10. A chip, characterized in that, The chip includes a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the steps of the ship departure time prediction method as described in any one of claims 1 to 6.

Images