Multi-target fusion operation fleet intelligent scheduling method and system

Abstract

The invention discloses a multi-target fusion operation fleet intelligent scheduling method and system. The method comprises the following steps: acquiring target voyage number plan data, historical and real-time voyage data of each ship in an operation fleet to be scheduled and the like; evaluating ships which can reach the target loading port and the target unloading port on time by using the constructed multi-factor navigational speed prediction model, and determining the ships as ETA on-time arrival ships; predicting a no-load stage fuel consumption range and full-load fuel consumption by utilizing the constructed multiple relation models in combination with different loading conditions and weather in a loading stage and an unloading stage, and performing CII index grade evaluation on each ETA punctual arrival ship to obtain a ship meeting a C-grade CII index; the TCE calculation model of the target voyage number is used to solve the TCE optimal solution of each ship meeting the C-level CII index, and the ship with the optimal TCE is determined as a scheduling ship to realize intelligent scheduling; and the multi-target evaluation result and intelligent scheduling information of the ETA index, the CII index and the TCE index are visually displayed so as to assist a decision maker in ship scheduling.

Description

Technical Field

[0001] The present invention relates to the field of ship control technology, and in particular to a multi-objective fusion intelligent scheduling method and system for an operating fleet. Background Art

[0002] With the continuous growth of global trade, ocean shipping plays an increasingly important role in international commerce. However, current fleet scheduling methods are constrained by multiple factors, including volatile weather conditions, ocean currents, and vessel capabilities. They often rely on empirical judgment and static data estimates, resulting in low accuracy and poor adaptability. Coupled with the strengthening of global shipping regulations and increasingly stringent international shipping emission reduction standards, the shipping industry is facing unprecedented regulatory challenges and tests.

[0003] At present, fleet scheduling mainly relies on decision makers to judge the optimal scheduling ships based on current static data and experience, which leads to problems such as poor transportation punctuality and unqualified carbon emissions. In addition, the scheduled ships are not economically optimal, affecting the market competitiveness of the operating fleet.

[0004] Therefore, establishing a ship scheduling model and decision-making support method that integrates multiple objectives such as timeliness indicators, emission indicators, and economic indicators is of vital importance for shipping companies to improve logistics efficiency, reduce transportation costs, and maintain market competitive advantages. Summary of the Invention

[0005] To address the low accuracy and adaptability of existing fleet scheduling methods, which rely on empirical judgment and static data estimation, this invention proposes a multi-objective integrated intelligent scheduling method for operating fleets. This method integrates multiple objectives, including timeliness, emissions, and economic indicators. This method ensures that fleet scheduling maximizes operational efficiency while meeting on-time arrival and emission requirements. This provides comprehensive decision-making support to decision makers, significantly enhancing the fleet's market competitiveness. The invention also relates to a multi-objective integrated intelligent scheduling system for operating fleets.

[0006] The present invention is achieved through the following technical solutions:

[0007] A multi-objective integrated intelligent scheduling method for an operating fleet includes the following steps:

[0008] Data collection step: collecting target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes port charges, current fuel prices, and current rates of the target loading port and target unloading port; the historical navigation data includes historical speeds, historical main engine speeds, historical fuel consumption, and corresponding historical meteorological information under different load conditions; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, first meteorological information between the current position of each ship and the target loading port, and second meteorological information between the target loading port and the target unloading port;

[0009] ETA on-time arrival evaluation steps: adopting machine learning algorithm, taking historical speed under different load conditions and corresponding historical meteorological information in historical navigation data as training samples, taking load condition and weather as input variables and speed as output variable, constructing a multi-factor speed prediction model; inputting the first meteorological information and empty state in real-time navigation data into the multi-factor speed prediction model, obtaining the predicted speed in the loading phase, calculating the theoretical speed of arriving at the target loading port on time based on the empty sailing distance from the ship to the target loading port and the target loading time, and setting the predicted speed in the loading phase greater than or equal to the theoretical speed. The ship with the highest speed is determined as an on-time loading ship; the second meteorological information and the full load status are input into a multi-factor speed prediction model to obtain a predicted speed for the unloading phase, and the ship with the predicted speed for the unloading phase greater than or equal to the contract speed is determined as an on-time unloading ship; the ship that meets both the requirements of the on-time loading ship and the on-time unloading ship is evaluated as an ETA on-time arrival ship; the minimum speed at which each ETA on-time arrival ship can reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute a first unloaded speed range;

[0010] CII index level evaluation steps: Based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a speed-main engine speed model under different meteorological and load conditions; and based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a main engine speed-fuel consumption model under different meteorological and load conditions; using the constructed no-load speed-main engine speed model and no-load main engine speed-fuel consumption model under different meteorological conditions, the first air velocity corresponding to each ETA arriving on time ship sailing at the first no-load speed range under the first meteorological information is generated. The full-load fuel consumption range is obtained; using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to each ETA on-time arrival ship sailing at the full-load speed under the second weather information is generated, where the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance, the CII index level of each ETA on-time arrival ship is evaluated to obtain ships that meet the C-level CII index; a second no-load fuel consumption range corresponding to the ships that meet the C-level CII index is determined, and a second no-load speed range is determined based on the second no-load fuel consumption range;

[0011] TCE indicator evaluation step: calculating the revenue of the target voyage based on the target cargo volume, the contract freight rate, and the current rate; using the TCE calculation model of the target voyage, based on the revenue of the target voyage, the port fee, and the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance, and the contract speed, solving the TCE optimal solution for each ship that meets the C-level CII indicator, and determining the ship with the optimal TCE as the scheduling ship to achieve intelligent scheduling;

[0012] Decision-making support output step: Visualization technology is used to intuitively display the multi-objective evaluation results and intelligent scheduling information of ETA indicators, CII indicators and TCE indicators, so that decision makers can schedule ships to the target voyage based on the scheduling information.

[0013] Preferably, in the data collection step, the collected historical navigation data and real-time navigation data are pre-processed, including eliminating abnormal data, performing ten-minute averaging, and screening data of the constant speed navigation phase and open water ship navigation data;

[0014] In the ETA on-time arrival evaluation step, the machine learning algorithm adopts the random forest algorithm, and optimizes the number of decision trees and the maximum depth through grid search, wherein the number of decision trees ranges from 50 to 200, and the maximum depth ranges from 3 to 10.

[0015] Preferably, in the CII index level evaluation step, a polynomial regression algorithm is used to construct ship speed-main engine speed models under different weather conditions and load conditions, including no-load ship speed-main engine speed models under different weather conditions and full-load ship speed-main engine speed models under different weather conditions; and a polynomial regression algorithm is used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load main engine speed-fuel consumption models under different weather conditions and full-load main engine speed-fuel consumption models under different weather conditions;

[0016] Generating, by using the constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at a first no-load speed range under the first weather information, including: generating, by using the constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA on-time arriving ship sailing at the first no-load speed range under the first weather information; and then generating, by combining the constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at each no-load main engine speed range under the first weather information;

[0017] Using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to the sailing at the full-load speed of each ship with an ETA arriving on time under the second weather information is generated, including: using the constructed full-load speed-main engine speed model under different weather conditions to generate the full-load main engine speed corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information; and then combining with the constructed full-load main engine speed-fuel consumption model under different weather conditions to generate the full-load fuel consumption corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information.

[0018] Preferably, in the CII index level evaluation step, the step of constructing the no-load speed-main engine speed model under different weather conditions is specifically as follows: under no-load conditions, using historical speed and historical weather information as independent variables and historical main engine speed as dependent variable, constructing a no-load speed-main engine speed regression model through feature transformation; using the pre-processed different historical weather information, historical speed and corresponding historical main engine speed under no-load conditions to fit the no-load speed-main engine speed regression model, estimating the regression coefficient, and forming the no-load speed-main engine speed model for each ship;

[0019] The steps for constructing the no-load main engine speed-fuel consumption model under different weather conditions are specifically as follows: under no-load conditions, using historical main engine speed and historical weather information as independent variables and historical fuel consumption as a dependent variable, constructing a no-load main engine speed-fuel consumption regression model through feature transformation; using the pre-processed different historical weather information, historical main engine speed, and corresponding historical fuel consumption under no-load conditions to fit the no-load main engine speed-fuel consumption regression model, estimating the regression coefficient, and forming the no-load main engine speed-fuel consumption model for each ship;

[0020] The steps of constructing the full-load speed-main engine speed model under different weather conditions are specifically as follows: under full-load conditions, using historical speed and historical weather information as independent variables and historical main engine speed as dependent variable, constructing a full-load speed-main engine speed regression model through feature transformation; using the pre-processed different historical weather information, historical speed and corresponding historical main engine speed under full-load conditions to fit the full-load speed-main engine speed regression model, estimating the regression coefficient, and forming a full-load speed-main engine speed model for each ship;

[0021] The steps for constructing the full-load main engine speed-fuel consumption model under different weather conditions are specifically as follows: under full-load conditions, using historical main engine speed and historical weather information as independent variables and historical fuel consumption as a dependent variable, constructing a full-load main engine speed-fuel consumption regression model through feature transformation; using preprocessed historical weather information, historical main engine speed, and corresponding historical fuel consumption under full-load conditions to fit the full-load main engine speed-fuel consumption regression model, estimating the regression coefficient, and forming a full-load main engine speed-fuel consumption model for each ship.

[0022] Preferably, in the CII index level assessment step, the CII index is calculated according to the CII calculation formula established by the International Maritime Organization based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance. The calculated CII index is compared with the annual CII rating threshold according to the annual CII benchmark value for each ship specified by the International Maritime Organization and the annual CII rating threshold set based on the CII benchmark value to determine the CII index level as A, B, C, D, or E, thereby screening out ships that meet the CII level CII index.

[0023] Preferably, the target voyage plan data in the data collection step further includes a loading operation time window and an unloading operation time window;

[0024] In the TCE indicator evaluation step, the TCE calculation model of the target voyage is: TCE = (target voyage income - port fees - fuel costs) ÷ sailing days, wherein fuel costs = full-load fuel consumption × (target full-load sailing distance ÷ contract speed) × current fuel price + second no-load fuel consumption range average × (no-load sailing distance ÷ second no-load speed range average) × current fuel price, sailing days = (no-load sailing distance ÷ second no-load speed range average) + loading operation time window + (target full-load sailing distance ÷ contract speed) + unloading operation time window.

[0025] Preferably, in the auxiliary decision output step, an interactive visual interface is used to display the multi-objective evaluation results and intelligent scheduling information, and the CII index distribution of each ship on the target voyage route is marked on the electronic nautical chart through a heat map. At the same time, the predicted trend of the TCE index with different meteorological changes is displayed in the form of a dynamic curve, providing decision makers with intuitive temporal and spatial dimension evaluation information;

[0026] Real-time data monitoring is also carried out. When the deviation between the real-time data and the target voyage plan data exceeds the preset threshold, the dynamic adjustment program of the scheduling plan is automatically triggered. By returning to the ETA on-time arrival evaluation step, the CII indicator level evaluation step and the TCE indicator evaluation step in sequence, the new scheduling ship is re-determined to achieve intelligent scheduling, and the adjustment process and results are output in a visual form.

[0027] A multi-objective integrated operating fleet intelligent dispatching system includes a data acquisition module, an ETA on-time arrival evaluation module, a CII index level evaluation module, a TCE index evaluation module and an auxiliary decision output module connected in sequence; wherein,

[0028] The data acquisition module is used to collect target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes port charges, current fuel prices, and current rates of the target loading port and target unloading port; the historical navigation data includes historical speeds, historical main engine speeds, historical fuel consumption, and corresponding historical meteorological information under different load conditions; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, first meteorological information between the current position of each ship and the target loading port, and second meteorological information between the target loading port and the target unloading port;

[0029] The ETA on-time arrival evaluation module is used to adopt a machine learning algorithm, take the historical speed under different load conditions and the corresponding historical meteorological information in the historical navigation data as training samples, take the load condition and weather as input variables, and the speed as output variable to build a multi-factor speed prediction model; input the first meteorological information and the empty state in the real-time navigation data into the multi-factor speed prediction model to obtain the loading stage predicted speed, calculate the theoretical speed of arriving at the target loading port on time based on the empty sailing distance from the ship to the target loading port and the target loading time, and set the loading stage predicted speed to be greater than or equal to the target speed. The ship having the theoretical speed is determined as an on-time loading ship; the second meteorological information and the full load status are input into a multi-factor speed prediction model to obtain a predicted speed for the unloading phase, and the ship having the predicted speed for the unloading phase greater than or equal to the contract speed is determined as an on-time unloading ship; the ship that meets both the requirements of the on-time loading ship and the on-time unloading ship is evaluated as an ETA on-time arrival ship; the minimum speed at which each ETA on-time arrival ship can reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute a first unloaded speed range;

[0030] The CII indicator level evaluation module is used to construct a ship speed-main engine speed model under different weather conditions and load conditions based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to construct a main engine speed-fuel consumption model under different weather conditions and load conditions based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to generate the first corresponding ETA corresponding to each ship arriving on time sailing at the first no-load speed range under the first meteorological information using the constructed no-load ship speed-main engine speed model and no-load main engine speed-fuel consumption model under different weather conditions. a no-load fuel consumption range; using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, generating the corresponding full-load fuel consumption of each ETA on-time arrival ship sailing at the full-load speed under the second weather information, where the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance, performing a CII index level assessment on each ETA on-time arrival ship to obtain ships that meet the CII index of Class C; determining a second no-load fuel consumption range corresponding to the ships that meet the CII index of Class C, and determining a second no-load speed range based on the second no-load fuel consumption range;

[0031] The TCE indicator evaluation module is configured to calculate the revenue of the target voyage based on the target cargo volume, the contract freight rate, and the current rate; and to utilize a TCE calculation model for the target voyage to determine the optimal TCE solution for each ship that meets the Class C CII indicator based on the revenue of the target voyage, the port fee, the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance, and the contract speed, and to determine the ship with the optimal TCE as the dispatching ship to achieve intelligent dispatching;

[0032] The auxiliary decision output module uses visualization technology to intuitively display the multi-objective evaluation results and intelligent scheduling information of the ETA index, CII index and TCE index, so that decision makers can schedule ships to the target voyage based on the scheduling information.

[0033] Preferably, the data collection module also pre-processes the collected historical navigation data and real-time navigation data, including eliminating abnormal data, performing ten-minute averaging, and screening constant speed navigation phase data and open water ship navigation data;

[0034] In the ETA on-time arrival assessment module, the machine learning algorithm adopts the random forest algorithm, and optimizes the number of decision trees and the maximum depth through grid search, wherein the number of decision trees ranges from 50 to 200, and the maximum depth ranges from 3 to 10.

[0035] Preferably, in the CII index level evaluation module, a polynomial regression algorithm is used to construct ship speed-main engine speed models under different weather conditions and load conditions, including no-load ship speed-main engine speed models under different weather conditions and full-load ship speed-main engine speed models under different weather conditions; a polynomial regression algorithm is used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load main engine speed-fuel consumption models under different weather conditions and full-load main engine speed-fuel consumption models under different weather conditions;

[0036] Generating, by using the constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at a first no-load speed range under the first weather information, including: generating, by using the constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA on-time arriving ship sailing at the first no-load speed range under the first weather information; and then generating, by combining the constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at each no-load main engine speed range under the first weather information;

[0037] Using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to the sailing at the full-load speed of each ship with an ETA arriving on time under the second weather information is generated, including: using the constructed full-load speed-main engine speed model under different weather conditions to generate the full-load main engine speed corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information; and then combining with the constructed full-load main engine speed-fuel consumption model under different weather conditions to generate the full-load fuel consumption corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information.

[0038] The beneficial effects of the present invention are as follows:

[0039] The present invention provides a multi-objective integrated operating fleet intelligent scheduling method. The method collects multi-source information such as target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the operating fleet to be scheduled, so as to provide a comprehensive and accurate data basis for subsequent scheduling. The method covers key data such as ship position, historical speed under different load conditions, historical main engine speed, historical fuel consumption and corresponding historical meteorological information, current fuel price, current rate and detailed target voyage plan data, thereby avoiding scheduling deviations caused by data loss and improving data reliability and integrity. The machine learning algorithm uses the historical speed under different load conditions in the historical navigation data and the corresponding historical meteorological information as training samples to establish a multi-factor speed prediction model. The speed of the empty and fully loaded stages (the predicted speed in the loading stage and the predicted speed in the unloading stage) is predicted based on real-time meteorological information. The speed is compared with the theoretical speed and the contract speed to determine whether the ship can arrive on time. Compared with the traditional method that relies on experience and static data to evaluate on-time arrival, the speed prediction model combined with meteorological data can avoid subjective experience errors and improve the credibility of the ship's punctual arrival. The dual verification mechanism of loading and unloading on time is fully The accuracy of ETA on-time arrival assessment is guaranteed; the minimum speed of each ETA on-time arrival ship that can reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute the first no-load speed range. The speed interval is delineated in combination with the minimum speed and the historical limit value, and the energy-saving potential is explored under time constraints to provide operating space for carbon emission optimization; the no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model are used to evaluate the fuel consumption in the no-load stage, and the full-load speed-main engine speed model and the full-load main engine speed-fuel consumption model are used to evaluate the fuel consumption in the full-load stage. Consumption is calculated, and the carbon intensity index level is assessed based on total fuel consumption. A meteorological impact model is introduced to dynamically correct fuel consumption forecasts to avoid misjudgments of carbon emissions caused by static data. A tiered filtering mechanism (C-level compliance) ensures compliance with the International Maritime Organization (IMO) environmental regulations and mitigates policy risks. The second no-load fuel consumption range corresponding to ships meeting the C-level carbon intensity index (CII index) is determined, and the second no-load speed range is determined based on the second no-load fuel consumption range. The speed range is inferred using environmental constraints to achieve a direct correlation between carbon emissions and speed, thereby defining the compliance boundary of speed for economic analysis.Based on the income of the target voyage, the port fee and the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance and the contract speed, the TCE calculation model of the target voyage is used to solve the TCE optimal solution of each ship that meets the C-level CII index, and the ship with the optimal TCE is determined as the scheduling ship to achieve intelligent scheduling. Based on multi-dimensional data, the TCE calculation model of the target voyage is used to quantify the economic benefits, avoid the loss of benefits caused by only considering a single CII index, and comprehensively consider the economic benefits. The present invention uses a triple coupling evaluation of timeliness, emission and economic indicators to break the contradiction between timeliness, environmental protection and economy in traditional scheduling and achieve Pareto optimality. The data-driven model replaces manual experience judgment to achieve efficient and intelligent scheduling of the operating fleet. The present invention assists in decision-making output through data collection, ETA on-time arrival assessment, CII index level assessment, TCE index assessment and visual display of the most efficient ship information, integrates multiple targets such as timeliness index, emission index and economic index, and combines multi-factor speed prediction model, no-load speed-main engine speed model, no-load main engine speed-fuel consumption model, full-load speed-main engine speed model, full-load main engine speed-fuel consumption model and TCE calculation model. Through the collaborative construction of multiple models, a complete calculation system is constructed, with the multi-factor speed prediction model as the basis, combined with the speed-main engine speed and main engine speed-fuel consumption model in no-load and full-load states, to accurately quantify the ship's navigation performance; on this basis, the timeliness, emission and economic indicators are integrated, and the TCE calculation model is fully relied on. This comprehensive assessment enables progressive and mutually supportive interaction between various models and indicators, ensuring fleet scheduling that maximizes operational efficiency while meeting on-time arrival and emission requirements. The precise linkage of multiple models improves the accuracy of vessel navigation data calculations, providing a reliable basis for scheduling decisions. The deep integration of multiple objectives balances timeliness, environmental protection, and economy, transforming fleet scheduling from single-dimensional optimization to systematic and comprehensive improvements, significantly enhancing the scientific nature and practicality of scheduling solutions. This multi-objective ship scheduling and decision-making support method, integrating timeliness, emissions, and economic indicators, helps shipping companies improve logistics efficiency, reduce transportation costs, and provide comprehensive decision-making support to decision makers, significantly enhancing the market competitiveness of fleets.

[0040] The present invention first pre-processes the historical meteorological information, historical main engine speed and corresponding historical fuel consumption in the historical navigation data under no-load and full-load conditions (such as eliminating abnormal data, ten-minute averaging, screening constant speed navigation stage data and open water ship navigation data) to eliminate the interference of abnormal values and redundant information, improve the reliability and consistency of the model input data, and avoid prediction distortion caused by data deviation; then, with the historical speed and historical meteorological information as independent variables and the historical main engine speed as the dependent variable, a speed-main engine speed regression model is constructed through feature transformation, and meteorological information (such as wind level, wind direction, flow rate) and speed are used as independent variables to capture the nonlinear influence of environmental factors on the main engine speed, breaking through the limitations of traditional single-variable models and significantly improving. The prediction accuracy under complex working conditions is improved. The coupling effect of ship speed and meteorological conditions is decoupled through feature transformation (such as quadratic and cubic polynomials), and the contribution of different environmental parameters to the main engine power demand is quantified. Finally, the ship speed-main engine speed regression model is fitted using the pre-processed historical meteorological information, historical speed and corresponding historical main engine speed under no-load and full-load conditions. The regression coefficient is accurately estimated to form the no-load speed-main engine speed model and the full-load speed-main engine speed model of each ship. The regression model fitted based on historical data can accurately quantify the influence coefficient of speed change and meteorological fluctuation on the main engine speed. An independent model is built for each ship to accommodate individual differences such as ship age, main engine model, and maintenance status, avoid energy efficiency evaluation deviation, and ensure that the prediction results are consistent with the actual performance.

[0041] The present invention first pre-processes the historical meteorological information, historical main engine speed and corresponding historical fuel consumption in the historical navigation data under no-load and full-load conditions (such as eliminating abnormal data such as abnormal fuel consumption records, ten-minute averaging, and screening constant speed navigation stage data and open water ship navigation data) to eliminate noise interference, ensure the authenticity and consistency of the model training data, and avoid fuel consumption prediction deviations caused by data quality problems; then, with the historical main engine speed and historical meteorological information as independent variables and the historical fuel consumption as the dependent variable, a main engine speed-fuel consumption regression model is constructed through feature transformation, and the main engine speed and meteorological information (such as wind level, wind direction, and flow rate) are used as input variables to quantify the combined impact of the main engine load and the external environment on fuel consumption, thereby solving the static limitations of the traditional single speed-fuel consumption model. The limitation of the proposed method is to significantly improve the fuel consumption prediction accuracy under complex working conditions, and reveal the nonlinear relationship between speed and fuel consumption through feature transformation (such as quadratic and cubic polynomials); finally, the no-load main engine speed-fuel consumption regression model is fitted using the pre-processed historical meteorological information, historical main engine speed and corresponding historical fuel consumption under no-load and full-load conditions, and the regression coefficient is accurately estimated to form the no-load main engine speed-fuel consumption model and full-load main engine speed-fuel consumption model of each ship. The regression model fitted based on historical data can accurately quantify the influence coefficient of main engine speed change and meteorological fluctuation on fuel consumption, independently build a model for each ship, and be compatible with personalized factors such as main engine model differences, equipment aging degree, maintenance level, etc., to avoid the distortion of energy efficiency evaluation caused by general models and ensure that the fuel consumption prediction results are highly consistent with the actual operating characteristics.

[0042] This system uses electronic chart heat maps to visually display the CII distribution of each ship on a target route, quickly identifying ships with qualified carbon emission ratings. Dynamic curves are used to display the predicted trend of TCE indicators as they change with weather conditions, quantifying the impact of environmental factors such as wind level and direction on economic returns. Emissions and economic indicators are displayed on a single interface, overcoming the limitations of traditional single-target decision-making and providing decision-makers with intuitive assessment information across time and space. Significant changes in real-time data automatically trigger a dynamic scheduling adjustment program, re-determining new scheduled ships based on efficiency, emissions, and economic indicators, achieving intelligent scheduling.

[0043] The present invention also relates to a multi-objective integrated operating fleet intelligent scheduling system, which corresponds to the above-mentioned multi-objective integrated operating fleet intelligent scheduling method, and can be understood as a system for realizing the above-mentioned multi-objective integrated operating fleet intelligent scheduling method, including a data acquisition module, an ETA on-time arrival evaluation module, a CII index level evaluation module, a TCE index evaluation module and an auxiliary decision output module. The modules work together and can flexibly respond to changes in sea conditions and emergencies and maintain efficient operations by integrating multiple objectives such as timeliness indicators, emission indicators and economic indicators and a dynamic scheduling mechanism; using visualization technology to intuitively display key data, effectively reflect the actual performance level of the ship, assist decision makers in making scientific decisions, and improve the accuracy and timeliness of decisions; through data integration systems and optimization algorithms, achieve comprehensive optimization of fleet operations, improve overall operational efficiency, and reduce resource waste; through emission index constraints, optimize the fuel consumption and carbon emissions of each ship in the operating fleet, achieve green shipping, comply with international and local environmental protection regulations, and promote sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of the multi-objective fusion operating fleet intelligent scheduling method of the present invention.

[0045] Figure 2 It is a key decision flow chart of the multi-objective fusion operating fleet intelligent scheduling method of the present invention.

[0046] Figure 3 This is a structural diagram of the multi-objective fusion operating fleet intelligent scheduling system of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be described below with reference to the accompanying drawings.

[0048] The present invention discloses a multi-objective fusion intelligent scheduling method for an operating fleet, which aims to solve the problems of low accuracy and poor adaptability caused by the existing fleet scheduling method relying on experience judgment and static data estimation. Through the fusion evaluation of multiple objectives such as ETA on-time arrival evaluation, CII index level evaluation and TCE index evaluation, the scheduling ship for the target voyage is determined efficiently and accurately from the operating fleet to be scheduled. Under the constraints of meeting the on-time arrival and emission requirements, the scheduling arrangement that maximizes the fleet operation efficiency can be achieved, providing decision makers with comprehensive decision support, thereby significantly improving the market competitiveness of the fleet. The method can be executed by a processor or an electronic device with processing capabilities, such as Figure 1 and Figure 2As shown in the figure, firstly, key data is collected, and from the fleet to be scheduled, a multi-factor speed prediction model is used in combination with real-time meteorological data to screen out ETA on-time arrival ships, and the first no-load speed range of each ship in the ETA on-time arrival ships sailing on the no-load sailing distance is determined. Secondly, the first no-load fuel consumption range of sailing at the first no-load speed range on the no-load sailing distance is determined, and the full-load fuel consumption of sailing at the contract speed on the target full-load sailing distance is determined. Afterwards, based on the first no-load fuel consumption range and the full-load fuel consumption, ships that meet the C-level carbon intensity index (CII index) are screened from the ETA on-time arrival ships. Then, the ships that meet the CII index are determined. The second no-load fuel consumption range of ships meeting the C-level carbon intensity index is determined based on the second no-load fuel consumption range, and the second no-load speed range is determined based on the second no-load speed range of each ship meeting the C-level carbon intensity index. Then, based on the second no-load speed range of each ship meeting the C-level carbon intensity index, the optimal economic benefit of each ship meeting the C-level carbon intensity index is determined, that is, the TCE index evaluation, and the TCE optimal solution of each ship meeting the C-level CII index is solved, and the ship with the best economic benefit is determined as the scheduling ship; finally, the scheduling information of the scheduling ship is intuitively displayed through visualization technology, so that decision makers can schedule ships to the target voyage based on the scheduling information. The present invention takes real-time meteorological data into consideration in the ETA on-time arrival evaluation, CII index level evaluation and TCE index evaluation, and can accurately realize the scheduling arrangement that maximizes the fleet operation efficiency. Specifically, the method includes the following steps:

[0049] 1. Data collection step: Collect target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein, the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes port charges, current fuel prices, and current rates of the target loading port and target unloading port; the historical navigation data includes historical speeds, historical main engine speeds, historical fuel consumption, and corresponding historical meteorological information under different load conditions; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, first meteorological information between the current position of each ship and the target loading port, and second meteorological information between the target loading port and the target unloading port.

[0050] Before intelligent scheduling, the embodiment of the present invention collects and stores relevant target voyage plan data, ship transportation market data and ship data, etc., wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, loading operation time window, unloading operation time window, contract speed, target full load sailing distance, contract freight rate, target cargo capacity, target voyage number, etc.; the ship transportation market data includes economic data such as port charges (such as usage fees, management fees, etc.) of the target loading port and target unloading port, current rates and current fuel prices; the ship data includes basic data of each ship in the fleet to be scheduled, equipment-related data, historical navigation data and real-time navigation data (ship-side data communication can be used to realize real-time collection, transmission and storage of ship-side data), etc., and the basic ship data includes ship name, IMO number, ship registration, ship type and size, ship flag, ship deadweight tonnage The ship's equipment-related data include the number of main engines, main engine model, main engine rated power, main engine rated speed, main engine fuel type, the number of auxiliary engines, auxiliary engine model, auxiliary engine rated power, auxiliary engine fuel type, etc. The ship's historical navigation data includes the historical speed (water speed), historical main engine speed, historical weather (including wind level, wind direction, etc.), historical fuel consumption, historical water depth, historical speed over ground, historical shaft power, historical bow direction, historical water flow direction and other data recorded in the historical navigation of each ship under different load conditions. The real-time navigation data includes the current position information of each ship in the fleet to be dispatched, the empty sailing distance between the ship and the target loading port (that is, the remaining empty voyage), the first weather information between the current position of the ship and the target loading port (including wind level, wind direction, etc.), and the second weather information from the target loading port to the target unloading port (including wind level, wind direction, etc.).

[0051] Preferably, in the data collection step, the collected historical navigation data and real-time navigation data are also pre-processed, including eliminating abnormal data, performing ten-minute averaging, and screening data of the constant speed navigation stage and open water ship navigation data, wherein the abnormal data are eliminated: the historical main engine speed is less than or equal to 0rpm, the historical shaft power is less than or equal to 0kW, the historical fuel consumption is less than or equal to 0t / day, and the historical ship heading, historical wind direction, real-time wind direction, and historical water flow direction are not within the range of [0°, 360°];

[0052] Perform ten-minute averaging on all collected data to improve the data's anti-interference ability;

[0053] In order to avoid data fluctuations during the acceleration and deceleration phases of the ship, the ship's constant speed phase was selected as the analysis object, that is, the historical main engine speed fluctuation was less than 1 rpm;

[0054] Screen open water ship navigation data, that is, the historical water depth is greater than 100m, the historical water speed is greater than 8Kn, the historical land speed is greater than 8Kn, and the absolute value of the historical water flow speed is less than or equal to 0.5Kn.

[0055] Second, the ETA on-time arrival assessment steps: Using a machine learning algorithm, the historical speeds under different load conditions and corresponding historical weather information from historical navigation data are used as training samples. Loading conditions and weather conditions are used as input variables, and speed is used as the output variable to construct a multi-factor speed prediction model. This model then calculates whether ships with different distributions can deliver cargo according to the agreed time and ensure on-time arrival. This process consists of two main stages: the first is from the ship's position (the current position of the ship to be dispatched) to the loading port, known as the loading phase; the second is from the loading port to the unloading port according to the target voyage plan, known as the unloading phase. Specifically, the first meteorological information and empty-load status in the real-time navigation data are input into the multi-factor speed prediction model to obtain the predicted speed in the loading phase, and the theoretical speed for arriving at the target loading port on time is calculated based on the empty-load sailing distance of the ship to the target loading port and the target loading time, and the ship whose predicted speed in the loading phase is greater than or equal to the theoretical speed is determined as the loading on-time ship; the second meteorological information and full-load status are input into the multi-factor speed prediction model to obtain the predicted speed in the unloading phase, and the ship whose predicted speed in the unloading phase is greater than or equal to the contract speed is determined as the unloading on-time ship; the ship that meets both the requirements of the loading on-time ship and the unloading on-time ship is evaluated as the ETA on-time arrival ship, such as Figure 2 The figure shows that the agreed arrival time is met; ships that do not meet the agreed arrival time are eliminated; the minimum speed of each ETA on-time ship that can reach the target loading port at the target loading time is determined, and the minimum speed of each ETA on-time ship and the highest historical speed constitute the first no-load speed range.

[0056] Preferably, the machine learning algorithm of the embodiment of the present invention can adopt a random forest algorithm, and optimize the parameters of the number of decision trees and the maximum depth through grid search, wherein the value range of the number of decision trees is 50-200, and the value range of the maximum depth is 3-10.

[0057] In the embodiment of the present invention, for the subsequent CII index level evaluation and TCE index evaluation, the minimum speed and the highest historical speed of the ship that can arrive at the target loading port at the target loading time according to each ETA on time are used to form a first empty speed range.

[0058] III. CII Index Level Assessment Steps: Based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a speed-main engine speed model under different weather conditions and load conditions; and based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a main engine speed-fuel consumption model under different weather conditions and load conditions to characterize the ship performance indicators; using the constructed no-load speed-main engine speed model and no-load main engine speed-fuel consumption model under different weather conditions, the on-time arrival (ETA) of each ship is generated. The first no-load fuel consumption range corresponding to the ship arriving on time with ETA sailing at the first no-load speed range under the first weather information is generated by using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, and the full-load fuel consumption corresponding to the ship arriving on time with ETA sailing at the full-load speed under the second weather information is generated, and the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption and the target full-load sailing distance, the CII index level of each ship arriving on time with ETA is evaluated to obtain the ship that meets the CII index of level C, such as Figure 2 The chart shows ships that meet the C-level rating; ships that do not meet the C-level rating are excluded; the second no-load fuel consumption range corresponding to ships that meet the C-level CII index is determined, and the second no-load speed range is determined based on the second no-load fuel consumption range.

[0059] The Carbon Intensity Indicator (CII) quantifies the amount of carbon dioxide emissions per unit of cargo or per unit distance traveled by a ship during transportation. It is a key indicator for assessing a ship's environmental performance and developing emission reduction measures. The International Maritime Organization (IMO) establishes an annual CII benchmark value and rating thresholds based on this benchmark. By comparing a ship's actual CII value with the annual CII rating thresholds, its environmental performance is classified into five levels: A (best), B (major superior), C (moderate), D (minor inferior), or E (inferior). Ensuring that ships achieve a C rating or above is a key environmental compliance goal.

[0060] In this embodiment of the present invention, based on the first no-load fuel consumption range, no-load sailing distance, full-load fuel consumption, and target full-load sailing distance, the CII index level of each ETA on-time arrival ship is evaluated. The specific steps for obtaining ships that meet the C-level CII index are as follows:

[0061] The carbon intensity of ships arriving on time at each ETA is calculated using the following equations (1), (2), (3) and (4):

[0062] The relation (1) is:

[0063]

[0064] Among them, for the second no-load fuel consumption range, the average value is used to participate in the total fuel consumption calculation;

[0065] The relationship (2) is:

[0066] Total CO2 emissions = Total fuel consumption × Fuel CO2 emission coefficient

[0067] The relation (3) is:

[0068] Transport workload = ship's deadweight tonnage × total sailing distance

[0069] The deadweight tonnage of the ship represents the weight carried by the ship, and the total sailing distance represents the sum of the empty sailing distance and the target fully loaded sailing distance;

[0070] The relation (4) is:

[0071]

[0072] The above equations (1), (2), (3) and (4) are used to construct a relationship model between carbon emission intensity and the first no-load fuel consumption range, and combined with the carbon intensity index rating standard, the ships that meet the C-level carbon intensity index are obtained.

[0073] Furthermore, based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct speed-main engine speed models under different weather conditions and load conditions, including no-load speed-main engine speed models under different weather conditions and full-load speed-main engine speed models under different weather conditions. Based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load speed-fuel consumption models under different weather conditions and full-load speed-fuel consumption models under different weather conditions. Using the pre-constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA arriving on time ship sailing at a first no-load speed range under the first weather information is generated. The specific steps are as follows: using the pre-constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA arriving on time ship sailing at the first no-load speed range under the first weather information is generated; and then combining the pre-constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA arriving on time ship sailing at each no-load main engine speed range under the first weather information is generated.

[0074] Furthermore, the full-load speed-main engine speed model and the full-load main engine speed-fuel consumption model under different weather conditions that are pre-constructed are used to generate the full-load fuel consumption corresponding to the full-load speed of each ship arriving on time with an ETA under the second weather information. The specific steps are as follows: the full-load speed-main engine speed corresponding to the full-load speed of each ship under the second weather information is generated by using the pre-constructed full-load speed-main engine speed model under different weather conditions; and the full-load fuel consumption corresponding to the full-load speed of each ship arriving on time with an ETA under the second weather information is generated in combination with the pre-constructed full-load main engine speed-fuel consumption model under different weather conditions.

[0075] This invention uses a polynomial regression algorithm to fit nonlinear data by adding higher-order terms of the independent variable. Polynomial regression has the advantages of being able to fit nonlinear data, being easy to implement, and being able to control the model complexity by adjusting the degree of the polynomial. The order of polynomial regression determines the complexity of the polynomial. A higher order results in a more complex model, allowing for more flexible data fitting, thus characterizing the nonlinear relationship between variables such as speed and fuel consumption.

[0076] Preferably, the steps for constructing the no-load speed-main engine speed model under different weather conditions in the embodiment of the present invention are specifically as follows:

[0077] Under no-load conditions, with historical ship speed and historical meteorological information as independent variables and historical main engine speed as the dependent variable, the following no-load speed-main engine speed regression model (quadratic polynomial) is constructed through feature transformation:

[0078]

[0079] Among them, x1 is the historical meteorological data, x2 is the historical speed data, y is the historical engine speed data, β0, β1, ...β n is the regression coefficient;

[0080] The following formula is used as the loss function, and the regression coefficient is calculated by the least squares method based on the pre-processed historical meteorological information, historical ship speed, and corresponding historical engine speed under no-load conditions:

[0081]

[0082] Among them, y i is the dependent variable, x i1 is the i-th historical meteorological data, x i2 is the i-th historical speed data, β0, β1, ...β n are the regression coefficients corresponding to the independent variables;

[0083] The smaller the value of the loss function, the more convergent the regression model training is. Ultimately, the regression coefficient is accurately estimated to form a no-load speed-main engine speed model for each ship under different weather conditions.

[0084] Preferably, the steps for constructing the no-load engine speed-fuel consumption model under different weather conditions in the embodiment of the present invention are specifically as follows:

[0085] Under no-load conditions, the historical main engine speed and historical meteorological information are used as independent variables, and the historical fuel consumption is used as the dependent variable. The no-load main engine speed-fuel consumption regression model (quadratic polynomial) is constructed through feature transformation:

[0086]

[0087] Among them, x1 is the historical meteorological data, x2 is the historical engine speed data, y is the historical fuel consumption data, β0, β1, ...β n is the regression coefficient;

[0088] The following formula is used as the loss function, and the regression coefficient is calculated by the least squares method based on the pre-processed historical meteorological information, historical engine speed, and corresponding historical fuel consumption under no-load conditions:

[0089]

[0090] Among them, y i is the dependent variable, x i1 is the i-th historical meteorological data, x i2 is the i-th historical engine speed data, β0, β1, ...β n are the regression coefficients corresponding to the independent variables;

[0091] The smaller the value of the loss function, the more convergent the regression model training is. Ultimately, the regression coefficient is accurately estimated to form the no-load main engine speed-fuel consumption model of each ship under different weather conditions.

[0092] Preferably, the steps for constructing the full-load speed-main engine speed model under different weather conditions in the embodiment of the present invention are specifically as follows:

[0093] Under full load conditions, the historical ship speed and historical meteorological information are used as independent variables, and the historical main engine speed is used as the dependent variable. A full load speed-main engine speed regression model (quadratic polynomial) is constructed through feature transformation:

[0094]

[0095] Among them, x1 is the historical meteorological data, x2 is the historical speed data, y is the historical main engine speed, β0, β1, ...β n is the regression coefficient;

[0096] The following formula is used as the loss function, and the regression coefficient is calculated by the least squares method based on the pre-processed historical meteorological information, historical ship speed, and corresponding historical main engine speed under full load conditions:

[0097]

[0098] Among them, y i is the dependent variable, x i1 is the i-th historical meteorological data, x i2 is the i-th historical speed data, β0, β1, ...β n are the regression coefficients corresponding to the independent variables;

[0099] The smaller the value of the loss function, the more convergent the regression model training is. Ultimately, the regression coefficient is accurately estimated to form a full load speed-main engine speed model for each ship under different weather conditions.

[0100] Preferably, the steps for constructing the full-load engine speed-fuel consumption model under different weather conditions in the embodiment of the present invention are specifically as follows:

[0101] Under full-load conditions, the historical main engine speed and historical meteorological information are used as independent variables, and the historical fuel consumption is used as the dependent variable. A full-load main engine speed-fuel consumption regression model (quadratic polynomial) is constructed through feature transformation:

[0102]

[0103] Among them, x1 is the historical meteorological data, x2 is the historical engine speed data, y is the historical fuel consumption data, β0, β1, ...β n is the regression coefficient;

[0104] The following formula is used as the loss function, and the regression coefficient is calculated by the least squares method based on the preprocessed historical meteorological information, historical engine speed, and corresponding historical fuel consumption under full load conditions:

[0105]

[0106] Among them, y i is the dependent variable, x i1 is the i-th historical meteorological data, x i2 is the i-th historical engine speed data, β0, β1, ...β n are the regression coefficients corresponding to the independent variables;

[0107] The smaller the value of the loss function, the more convergent the regression model training is. Ultimately, the regression coefficient is accurately estimated to form a full-load main engine speed-fuel consumption model for each ship under different weather conditions.

[0108] IV. TCE Index Evaluation Step: Calculate the target voyage's revenue based on the target cargo capacity, contracted freight rate, and current rates. Utilize the target voyage's TCE calculation model to determine the optimal TCE solution for each vessel that meets the C-level CII indicators based on the target voyage's revenue, port fees, current fuel prices, the full-load fuel consumption, the second light-load fuel consumption range, the second light-load speed range, the light-load sailing distance, the target full-load sailing distance, and the contracted speed. The vessel with the optimal TCE is then designated as the dispatch vessel for intelligent scheduling. This step analyzes vessel operational motivation based on meeting the CII indicators from the CII Index Level Evaluation Step. By comprehensively considering fuel costs, port fees, and other operating costs, the economic benefits of different vessels are evaluated, and the sailing plan with the greatest daily benefits is selected.

[0109] The embodiment of the present invention uses the following relationship to calculate the revenue of the target voyage:

[0110] C 收入 =MT×WS×freight rate

[0111] Among them, C 收入 represents the revenue of the target voyage, MT represents the target cargo capacity, WS represents the rate (which is the benchmark freight index in the tanker shipping market and provides a unified freight calculation standard for global shipping), and freight rate represents the contract freight rate.

[0112] The TCE calculation model constructed in the embodiment of the present invention is as follows:

[0113]

[0114] C 成本 =Port fee + management fee + fuel fee + other

[0115]

[0116] Among them, C 收入 Indicates the revenue of the target voyage; C 成本 represents the cost of the target voyage; others represent consulting fees, agency fees, crew wages, and other expenses; sailing days represents the total time it takes for the vessel to sail from its current position to the target unloading port and complete cargo unloading; the full-load fuel consumption and the second empty-load fuel consumption ranges are both obtained in the CII indicator level assessment step. For the second empty-load fuel consumption range, the average value is used in the TCE calculation;

[0117] Finally, the optimal economic benefit of each ship that meets the C-level CII index, that is, the optimal TCE solution, is solved, and the ship with the optimal CII, that is, the optimal economic benefit, is determined as the scheduling ship to achieve intelligent scheduling.

[0118] 5. Decision-making support output step: Visualization technology is used to intuitively display the multi-objective evaluation results of the ETA index, CII index, and TCE index, as well as intelligent scheduling information, so that decision makers can schedule ships to the target voyage based on the scheduling information.

[0119] The embodiment of the present invention uses an interactive visualization interface to display multi-objective assessment results and intelligent scheduling information. For example, through charts and dashboards, it can intuitively display the fleet's position, navigation path, estimated arrival time, carbon intensity index and economic analysis results. It also uses heat maps to mark the distribution of CII indicators of each ship on the target voyage route on the electronic nautical chart. At the same time, it displays the predicted trend of TCE indicators with different meteorological changes in the form of dynamic curves, providing decision makers with intuitive temporal and spatial dimension assessment information, thereby improving the accuracy and timeliness of decision-making.

[0120] Furthermore, the embodiment of the present invention also performs real-time data monitoring. When the deviation between the real-time data and the target voyage plan data exceeds a preset threshold, the scheduling plan dynamic adjustment program is automatically triggered. By returning to the ETA on-time arrival evaluation step, the CII indicator level evaluation step and the TCE indicator evaluation step in sequence, a new scheduling ship is re-determined to achieve intelligent scheduling, and the adjustment process and results are output in a visual form. Through real-time monitoring and dynamic adjustment, real-time data is combined with historical data analysis and corresponding prediction models to optimize scheduling decisions and achieve the optimization of each goal.

[0121] Based on the same inventive concept, one or more embodiments of this specification also provide a multi-objective integrated intelligent scheduling system for an operating fleet. Since the principles of the problems solved by the multi-objective integrated intelligent scheduling system for an operating fleet are similar to those of the aforementioned multi-objective integrated intelligent scheduling method for an operating fleet, the implementation of the multi-objective integrated intelligent scheduling system for an operating fleet can refer to the aforementioned implementation of the multi-objective integrated intelligent scheduling method for an operating fleet, and the repeated parts will not be repeated.

[0122] Figure 3 This is a structural diagram of a multi-objective integrated fleet intelligent scheduling system provided in one or more embodiments of this specification. Figure 3 As shown, the multi-objective fusion operating fleet intelligent scheduling system includes a data acquisition module 101, an ETA on-time arrival evaluation module 102, a CII index level evaluation module 103, a TCE index evaluation module 104 and an auxiliary decision output module 105.

[0123] The data acquisition module 101 is used to collect target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes the port charges of the target loading port and target unloading port, the current fuel price and the current rate; the historical navigation data includes the historical speed under different load conditions, the historical main engine speed, the historical fuel consumption and the corresponding historical meteorological information; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, the first meteorological information between the current position of each ship and the target loading port, and the second meteorological information from the target loading port to the target unloading port.

[0124] The ETA on-time arrival assessment module 102 is used to adopt a machine learning algorithm, take the historical speeds under different load conditions and the corresponding historical meteorological information in the historical navigation data as training samples, take the load condition and weather as input variables, and the speed as output variable, to build a multi-factor speed prediction model; input the first meteorological information and the empty state in the real-time navigation data into the multi-factor speed prediction model to obtain the predicted speed in the loading phase, calculate the theoretical speed of arriving at the target loading port on time based on the empty sailing distance from the ship to the target loading port and the target loading time, and set the predicted speed in the loading phase to be greater than or equal to the target speed. The ship with the theoretical speed is determined as the loading on-time ship; the second meteorological information and the full load status are input into the multi-factor speed prediction model to obtain the predicted speed in the unloading stage, and the ship with the predicted speed in the unloading stage greater than or equal to the contract speed is determined as the unloading on-time ship; the ship that meets both the requirements of the loading on-time ship and the unloading on-time ship is evaluated as the ETA on-time arrival ship; the minimum speed of each ETA on-time arrival ship to reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute the first no-load speed range.

[0125] The CII indicator level evaluation module 103 is used to construct a ship speed-main engine speed model under different weather conditions and load conditions based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to construct a main engine speed-fuel consumption model under different weather conditions and load conditions based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to generate the first corresponding equations for each ETA arriving on time when the ship sails in the first no-load speed range under the first meteorological information using the constructed no-load ship speed-main engine speed model and no-load main engine speed-fuel consumption model under different weather conditions. a no-load fuel consumption range; using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, generating the full-load fuel consumption corresponding to each ETA on-time arrival ship sailing at the full-load speed under the second weather information, where the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance, performing a CII index level assessment on each ETA on-time arrival ship to obtain ships meeting the CII index of Class C; determining a second no-load fuel consumption range corresponding to the ships meeting the CII index, and determining a second no-load speed range based on the second no-load fuel consumption range.

[0126] The TCE indicator evaluation module 104 is used to calculate the revenue of the target voyage based on the target cargo volume, the contract freight rate and the current rate; based on the revenue of the target voyage, the port fee and the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance and the contract speed, using the TCE calculation model of the target voyage, solve the TCE optimal solution for each ship that meets the Class C CII indicator, and determine the ship with the optimal TCE as the scheduling ship to achieve intelligent scheduling.

[0127] The auxiliary decision output module 105 uses visualization technology to intuitively display the multi-objective evaluation results and intelligent scheduling information of the ETA index, CII index and TCE index, so that decision makers can schedule ships to the target voyage based on the scheduling information.

[0128] Furthermore, the data collection module 101 also pre-processes the collected historical navigation data and real-time navigation data, including eliminating abnormal data, ten-minute averaging, and screening constant speed navigation stage data and open water ship navigation data.

[0129] Furthermore, in the ETA on-time arrival assessment module 102, the machine learning algorithm adopts a random forest algorithm, and optimizes the number of decision trees and the maximum depth through grid search, wherein the number of decision trees ranges from 50 to 200, and the maximum depth ranges from 3 to 10.

[0130] Furthermore, in the CII index level evaluation module 103, a polynomial regression algorithm is used to construct ship speed-main engine speed models under different weather conditions and load conditions, including no-load ship speed-main engine speed models under different weather conditions and full-load ship speed-main engine speed models under different weather conditions. A polynomial regression algorithm is also used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load main engine speed-fuel consumption models under different weather conditions and full-load main engine speed-fuel consumption models under different weather conditions.

[0131] Generating, by using the constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at a first no-load speed range under the first weather information, including: generating, by using the constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA on-time arriving ship sailing at the first no-load speed range under the first weather information; and then generating, by combining the constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at each no-load main engine speed range under the first weather information;

[0132] Using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to the sailing at the full-load speed of each ship with an ETA arriving on time under the second weather information is generated, including: using the constructed full-load speed-main engine speed model under different weather conditions to generate the full-load main engine speed corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information; and then combining with the constructed full-load main engine speed-fuel consumption model under different weather conditions to generate the full-load fuel consumption corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information.

[0133] The present invention integrates ETA on-time arrival prediction, CII calculation and TCE economic analysis to develop an intelligent scheduling method and system for operating fleets. By integrating ship navigation data, fuel consumption data and performance analysis, and adopting a multi-objective calculation and prediction model, it ensures that the fleet can achieve optimal scheduling arrangements that maximize benefits while meeting the operating conditions of on-time arrival and emission requirements, providing comprehensive support for decision makers and significantly improving ship operating efficiency and market competitiveness.

[0134] It should be noted that the specific embodiments described above can enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although this specification has described the present invention in detail with reference to the drawings and embodiments, those skilled in the art should understand that the present invention can still be modified or replaced with equivalents. In short, all technical solutions and improvements that do not depart from the spirit and scope of the present invention should be included in the scope of protection of the patent for the present invention.

Claims

1. A multi-objective integrated intelligent scheduling method for operating fleets, characterized by: The following steps are involved: Data collection step: collecting target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes port charges, current fuel prices, and current rates of the target loading port and target unloading port; the historical navigation data includes historical speeds, historical main engine speeds, historical fuel consumption, and corresponding historical meteorological information under different load conditions; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, first meteorological information between the current position of each ship and the target loading port, and second meteorological information between the target loading port and the target unloading port; ETA on-time arrival evaluation steps: adopting machine learning algorithm, taking historical speed under different load conditions and corresponding historical meteorological information in historical navigation data as training samples, taking load condition and weather as input variables and speed as output variable, constructing a multi-factor speed prediction model; inputting the first meteorological information and empty state in real-time navigation data into the multi-factor speed prediction model, obtaining the predicted speed in the loading phase, calculating the theoretical speed of arriving at the target loading port on time based on the empty sailing distance from the ship to the target loading port and the target loading time, and setting the predicted speed in the loading phase greater than or equal to the theoretical speed. The ship with the highest speed is determined as an on-time loading ship; the second meteorological information and the full load status are input into a multi-factor speed prediction model to obtain a predicted speed for the unloading phase, and the ship with the predicted speed for the unloading phase greater than or equal to the contract speed is determined as an on-time unloading ship; the ship that meets both the requirements of the on-time loading ship and the on-time unloading ship is evaluated as an ETA on-time arrival ship; the minimum speed at which each ETA on-time arrival ship can reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute a first unloaded speed range; CII index level evaluation steps: Based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a speed-main engine speed model under different meteorological and load conditions; and based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data, a polynomial regression algorithm is used to construct a main engine speed-fuel consumption model under different meteorological and load conditions; using the constructed no-load speed-main engine speed model and no-load main engine speed-fuel consumption model under different meteorological conditions, the first air velocity corresponding to each ETA arriving on time ship sailing at the first no-load speed range under the first meteorological information is generated. The full-load fuel consumption range is obtained; using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to each ETA on-time arrival ship sailing at the full-load speed under the second weather information is generated, where the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance, the CII index level of each ETA on-time arrival ship is evaluated to obtain ships that meet the C-level CII index; a second no-load fuel consumption range corresponding to the ships that meet the C-level CII index is determined, and a second no-load speed range is determined based on the second no-load fuel consumption range; TCE indicator evaluation step: calculating the revenue of the target voyage based on the target cargo volume, the contract freight rate, and the current rate; using the TCE calculation model of the target voyage, based on the revenue of the target voyage, the port fee, and the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance, and the contract speed, solving the TCE optimal solution for each ship that meets the C-level CII indicator, and determining the ship with the optimal TCE as the scheduling ship to achieve intelligent scheduling; Decision-making support output step: Visualization technology is used to intuitively display the multi-objective evaluation results and intelligent scheduling information of ETA indicators, CII indicators and TCE indicators, so that decision makers can schedule ships to the target voyage based on the scheduling information.

2. The method according to claim 1, characterized in that In the data collection step, the collected historical navigation data and real-time navigation data are also pre-processed, including eliminating abnormal data, averaging over ten minutes, and screening data from the constant speed navigation phase and open water vessel navigation data; In the ETA on-time arrival evaluation step, the machine learning algorithm adopts the random forest algorithm, and optimizes the number of decision trees and the maximum depth through grid search, wherein the number of decision trees ranges from 50 to 200, and the maximum depth ranges from 3 to 10.

3. The method according to claim 2, characterized in that In the CII index level evaluation step, a polynomial regression algorithm is used to construct ship speed-main engine speed models under different weather conditions and load conditions, including no-load ship speed-main engine speed models under different weather conditions and full-load ship speed-main engine speed models under different weather conditions; a polynomial regression algorithm is used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load main engine speed-fuel consumption models under different weather conditions and full-load main engine speed-fuel consumption models under different weather conditions; Generating, by using the constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at a first no-load speed range under the first weather information, including: generating, by using the constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA on-time arriving ship sailing at the first no-load speed range under the first weather information; and then generating, by combining the constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at each no-load main engine speed range under the first weather information; Using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to the sailing at the full-load speed of each ship with an ETA arriving on time under the second weather information is generated, including: using the constructed full-load speed-main engine speed model under different weather conditions to generate the full-load main engine speed corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information; and then combining with the constructed full-load main engine speed-fuel consumption model under different weather conditions to generate the full-load fuel consumption corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information.

4. The method according to claim 3, characterized in that In the CII index level evaluation step, the step of constructing the no-load speed-main engine speed model under different weather conditions is specifically as follows: under no-load conditions, using historical speed and historical weather information as independent variables and historical main engine speed as dependent variable, constructing a no-load speed-main engine speed regression model through feature transformation; using the pre-processed different historical weather information under no-load conditions, historical speed and corresponding historical main engine speed to fit the no-load speed-main engine speed regression model, estimating the regression coefficient, and forming the no-load speed-main engine speed model for each ship; The steps for constructing the no-load main engine speed-fuel consumption model under different weather conditions are specifically as follows: under no-load conditions, using historical main engine speed and historical weather information as independent variables and historical fuel consumption as a dependent variable, constructing a no-load main engine speed-fuel consumption regression model through feature transformation; using the pre-processed different historical weather information, historical main engine speed, and corresponding historical fuel consumption under no-load conditions to fit the no-load main engine speed-fuel consumption regression model, estimating the regression coefficient, and forming the no-load main engine speed-fuel consumption model for each ship; The steps of constructing the full-load speed-main engine speed model under different weather conditions are specifically as follows: under full-load conditions, using historical speed and historical weather information as independent variables and historical main engine speed as dependent variable, constructing a full-load speed-main engine speed regression model through feature transformation; using the pre-processed different historical weather information, historical speed and corresponding historical main engine speed under full-load conditions to fit the full-load speed-main engine speed regression model, estimating the regression coefficient, and forming a full-load speed-main engine speed model for each ship; The steps for constructing the full-load main engine speed-fuel consumption model under different weather conditions are specifically as follows: under full-load conditions, using historical main engine speed and historical weather information as independent variables and historical fuel consumption as a dependent variable, constructing a full-load main engine speed-fuel consumption regression model through feature transformation; using preprocessed historical weather information, historical main engine speed, and corresponding historical fuel consumption under full-load conditions to fit the full-load main engine speed-fuel consumption regression model, estimating the regression coefficient, and forming a full-load main engine speed-fuel consumption model for each ship.

5. The method according to any one of claims 1 to 4, characterized in that In the CII index level assessment step, the CII index is calculated according to the CII calculation formula established by the International Maritime Organization based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance. The calculated CII index is compared with the annual CII rating threshold based on the annual CII benchmark value for each ship specified by the International Maritime Organization and the annual CII rating threshold set based on the CII benchmark value to determine the CII index level as A, B, C, D, or E, thereby screening out ships that meet the CII level CII index.

6. The method according to any one of claims 1 to 4, characterized in that The target voyage plan data in the data collection step also includes a loading operation time window and an unloading operation time window; In the TCE indicator evaluation step, the TCE calculation model of the target voyage is: TCE = (target voyage income - port fees - fuel costs) ÷ sailing days, wherein fuel costs = full-load fuel consumption × (target full-load sailing distance ÷ contract speed) × current fuel price + second no-load fuel consumption range average × (no-load sailing distance ÷ second no-load speed range average) × current fuel price, sailing days = (no-load sailing distance ÷ second no-load speed range average) + loading operation time window + (target full-load sailing distance ÷ contract speed) + unloading operation time window.

7. The method according to any one of claims 1 to 4, characterized in that In the decision-making support output step, an interactive visual interface is used to display the multi-objective assessment results and intelligent scheduling information. The CII index distribution of each ship along the target voyage route is marked on the electronic nautical chart through a heat map. At the same time, the predicted trend of the TCE index with different weather changes is displayed in the form of a dynamic curve, providing decision makers with intuitive temporal and spatial evaluation information. Real-time data monitoring is also carried out. When the deviation between the real-time data and the target voyage plan data exceeds the preset threshold, the dynamic adjustment program of the scheduling plan is automatically triggered. By returning to the ETA on-time arrival evaluation step, the CII indicator level evaluation step and the TCE indicator evaluation step in sequence, the new scheduling ship is re-determined to achieve intelligent scheduling, and the adjustment process and results are output in a visual form.

8. A multi-objective integrated fleet intelligent dispatching system, characterized by: It includes a data acquisition module, an ETA on-time arrival assessment module, a CII indicator level assessment module, a TCE indicator assessment module and an auxiliary decision output module connected in sequence; among which, The data acquisition module is used to collect target voyage plan data, ship transportation market data, and historical navigation data and real-time navigation data of each ship in the fleet to be scheduled; wherein the target voyage plan data includes the target loading port, target unloading port, target loading time, target unloading time, contract speed, target full-load sailing distance, contract freight rate, and target cargo capacity; the ship transportation market data includes port charges, current fuel prices, and current rates of the target loading port and target unloading port; the historical navigation data includes historical speeds, historical main engine speeds, historical fuel consumption, and corresponding historical meteorological information under different load conditions; the real-time navigation data includes the empty-load sailing distance between each ship in the fleet to be scheduled and the target loading port, first meteorological information between the current position of each ship and the target loading port, and second meteorological information between the target loading port and the target unloading port; The ETA on-time arrival evaluation module is used to adopt a machine learning algorithm, take the historical speed under different load conditions and the corresponding historical meteorological information in the historical navigation data as training samples, take the load condition and weather as input variables, and the speed as output variable to build a multi-factor speed prediction model; input the first meteorological information and the empty state in the real-time navigation data into the multi-factor speed prediction model to obtain the loading stage predicted speed, calculate the theoretical speed of arriving at the target loading port on time based on the empty sailing distance from the ship to the target loading port and the target loading time, and set the loading stage predicted speed to be greater than or equal to the target speed. The ship having the theoretical speed is determined as an on-time loading ship; the second meteorological information and the full load status are input into a multi-factor speed prediction model to obtain a predicted speed for the unloading phase, and the ship having the predicted speed for the unloading phase greater than or equal to the contract speed is determined as an on-time unloading ship; the ship that meets both the requirements of the on-time loading ship and the on-time unloading ship is evaluated as an ETA on-time arrival ship; the minimum speed at which each ETA on-time arrival ship can reach the target loading port at the target loading time is determined, and the minimum speed and the highest historical speed of each ETA on-time arrival ship constitute a first unloaded speed range; The CII indicator level evaluation module is used to construct a ship speed-main engine speed model under different weather conditions and load conditions based on the historical speed, historical main engine speed and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to construct a main engine speed-fuel consumption model under different weather conditions and load conditions based on the historical main engine speed, historical fuel consumption and corresponding historical meteorological information under different load conditions in the historical navigation data using a polynomial regression algorithm; and to generate the first corresponding ETA corresponding to each ship arriving on time sailing at the first no-load speed range under the first meteorological information using the constructed no-load ship speed-main engine speed model and no-load main engine speed-fuel consumption model under different weather conditions. a no-load fuel consumption range; using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, generating the corresponding full-load fuel consumption of each ETA on-time arrival ship sailing at the full-load speed under the second weather information, where the full-load speed is the contract speed; then, based on the first no-load fuel consumption range, the no-load sailing distance, the full-load fuel consumption, and the target full-load sailing distance, performing a CII index level assessment on each ETA on-time arrival ship to obtain ships that meet the CII index of Class C; determining a second no-load fuel consumption range corresponding to the ships that meet the CII index of Class C, and determining a second no-load speed range based on the second no-load fuel consumption range; The TCE indicator evaluation module is configured to calculate the revenue of the target voyage based on the target cargo volume, the contract freight rate, and the current rate; and to utilize a TCE calculation model for the target voyage to determine the optimal TCE solution for each ship that meets the Class C CII indicator based on the revenue of the target voyage, the port fee, the current fuel price, the full-load fuel consumption, the second no-load fuel consumption range, the second no-load speed range, the no-load sailing distance, the target full-load sailing distance, and the contract speed, and to determine the ship with the optimal TCE as the dispatching ship to achieve intelligent dispatching; The auxiliary decision output module uses visualization technology to intuitively display the multi-objective evaluation results and intelligent scheduling information of the ETA index, CII index and TCE index, so that decision makers can schedule ships to the target voyage based on the scheduling information.

9. The system according to claim 8, characterized in that The data acquisition module also pre-processes the collected historical navigation data and real-time navigation data, including eliminating abnormal data, averaging over ten minutes, and screening data from the constant speed navigation phase and open water vessel navigation data; In the ETA on-time arrival assessment module, the machine learning algorithm adopts the random forest algorithm, and optimizes the number of decision trees and the maximum depth through grid search, wherein the number of decision trees ranges from 50 to 200, and the maximum depth ranges from 3 to 10.

10. The system according to claim 8, wherein: In the CII index level evaluation module, a polynomial regression algorithm is used to construct ship speed-main engine speed models under different weather conditions and load conditions, including no-load ship speed-main engine speed models under different weather conditions and full-load ship speed-main engine speed models under different weather conditions. A polynomial regression algorithm is used to construct main engine speed-fuel consumption models under different weather conditions and load conditions, including no-load main engine speed-fuel consumption models under different weather conditions and full-load main engine speed-fuel consumption models under different weather conditions. Generating, by using the constructed no-load speed-main engine speed model and the no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at a first no-load speed range under the first weather information, including: generating, by using the constructed no-load speed-main engine speed model under different weather conditions, a no-load main engine speed range corresponding to each ETA on-time arriving ship sailing at the first no-load speed range under the first weather information; and then generating, by combining the constructed no-load main engine speed-fuel consumption model under different weather conditions, a first no-load fuel consumption range corresponding to each ETA on-time arriving ship sailing at each no-load main engine speed range under the first weather information; Using the constructed full-load speed-main engine speed model and full-load main engine speed-fuel consumption model under different weather conditions, the full-load fuel consumption corresponding to the sailing at the full-load speed of each ship with an ETA arriving on time under the second weather information is generated, including: using the constructed full-load speed-main engine speed model under different weather conditions to generate the full-load main engine speed corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information; and then combining with the constructed full-load main engine speed-fuel consumption model under different weather conditions to generate the full-load fuel consumption corresponding to the full-load speed of each ship with an ETA arriving on time under the second weather information.

Images