Data-driven fuel cabin planner system

Abstract

A system (11) for a data-driven fuel tank planner for optimizing the refueling cost of a marine vessel (1) is described. A system (11) of a fuel tank planner is operable on a computing system (10) located on a vessel (1) and includes a data acquisition module (111), a fuel price prediction module (112), a data optimization module (113). A fuel price prediction module (112) predicts a long-term or short-term future fuel price and / or fuel index. A data optimization module (113) determines an optimal time for fulfilling a refueling clause contract or combination of contracts based on the predicted fuel price and / or fuel index. A predicted fuel price and / or fuel index, an optimized fuel clause contract, and a fueling plan are output by a notification generation module (114) and a human-machine interface system (18). The output of the system of the fuel tank planner is validated using real data trained on various customized machine learning and optimization models.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] The present invention claims priority from Singapore patent application no.10202203430W filed on April 4, 2022, the entire disclosure of which is incorporated herein by reference. Technical Field

[0003] The present invention relates to refueling of marine vessels. More specifically, the present invention relates to a system of data-driven bunker planners that optimizes bunkering costs by predicting fuel prices and optimizing data related to planned routes, vessel fuel consumption, vessel fuel storage restrictions (constraints) and bunkering facilities at ports of call. Background Art

[0004] The refueling of a ship is the process of replenishing fuel for a ship to continue its planned functions. The refueling of a ship in sea travel is called bunkering. Such ships can only store a limited amount of fuel on board, so they must plan and call at various ports during their refueling journey.

[0005] Since not every port provides bunkering facilities, the crew of the ship needs to plan the purchase of fuel before or during the ship's journey so that the ship has enough fuel to complete the journey. Preferably, in order to minimize the operating costs of the ship, the crew must consider the price of fuel when refueling at the selected bunkering port or when signing a bunkering contract with the bunkering supplier.

[0006] However, the price of fuel may vary at each bunkering port. In addition to this, fuel prices may fluctuate from day to day during the vessel’s journey. As a result, it may be difficult for the vessel’s crew to optimally plan the vessel’s bunkering operations. Depending on the fuel price of the day, it may be best to consider an alternative route or designate a bunkering contract to be fulfilled, thereby optimizing the cost of the vessel’s bunkering operations.

[0007] There are many known systems that may be related to managing the cost of refueling a ship. For example, U.S. Patent No. 10,139,235 assigned to Nippon Yusen describes a data processing device that generates display instruction data for a specific fuel price and a determination rule for the amount of refueling at one or more refueling locations, and outputs the data to a terminal device that displays information for formulating a refueling plan. The data processing includes route identification, reference fuel price data, fuel price data, price determination rules, refueling planning data, refueling locations, remaining oil data, refueling amounts, rules for each route, and rules for each port. The complex task of formulating a ship refueling plan is simplified by the presence of this data processing device.

[0008] Another known technology is U.S. Publication No. 20160265920 assigned to NYK Line Co., Ltd., which provides a refueling plan that can reduce the fuel cost of a ship during voyage. It also provides an apparatus including a collection unit for fuel consumption, port-to-port distance, voyage plan, remaining fuel, fuel price, refueling plan, fuel cost calculation, and output notification related to fuel cost, thereby providing a fuel supply plan.

[0009] However, it should be noted that US Patent No. 10139235 does not guarantee the quality and performance of pre-specified decision rules in terms of fuel tank cost optimization. However, US20160265920 only generates a feasible bunkering plan, not an optimal bunkering plan that minimizes expected future fuel costs. Therefore, it is desirable to have a data-driven invention to optimize bunkering costs when a ship moves from one port to another according to a predetermined route; preferably, some data involves predictions of fuel prices and / or fuel indices, and determining the optimal time, location and quantity for fulfilling a bunkering contract based on the predicted fuel prices or fuel indices. Summary of the invention

[0010] The object of the present invention is to seek a data-driven bunker planner (bunkerplanner) system, which is used to optimize the refueling cost of ships at sea. In order to achieve this purpose, the present invention sets a computing system and a bunker planner system (a bunker planner system, a bunker planning system) on the ship. More specifically, the bunker planner system also includes a fuel price prediction module and a data optimization module. The prediction module is used to predict the future long-term or short-term fuel price and / or fuel index. The data optimization module is used to determine the best route based on the predicted fuel price and / or fuel index, and to determine the best time to fulfill one or more refueling terms contracts based on the predicted fuel price and / or fuel index.

[0011] Advantageously, the present invention uses a data-driven, non-parametric approach to support decision makers in a dynamic setting. Furthermore, the use of forecasting and data optimization enables decision makers to determine the most economical amount of fuel to purchase and results in appropriate fuel purchase planning with the lowest cost. Furthermore, the data-driven bunker planner system is compatible with today's ship computing systems for implementation or integration.

[0012] In one embodiment, the present invention provides a system of data-driven bunker planner for minimizing the bunkering cost of a ship. The system of data-driven bunker planner predicts fuel prices and / or fuel indices and optimizes data related to planned routes, fuel consumption, vessel restrictions, and ports of call.

[0013] Preferably, the bunker planner system further determines the optimal time, quantity and location of refueling to fulfill one or more refueling term contracts based on predicted fuel prices and / or fuel indices.

[0014] Preferably, the system of the fuel tank planner also includes a data acquisition module.

[0015] Preferably, the data collection module also includes a refueling terms contract data collection submodule; a feature data collection submodule; a fuel data collection submodule; a route data collection submodule; a fuel consumption data collection submodule; a ship status data collection submodule; and a port data collection submodule.

[0016] Preferably, the fuel tank planner system further comprises a fuel price prediction module.

[0017] Preferably, the fuel price prediction module further includes a long-term prediction submodule and a short-term prediction submodule.

[0018] Preferably, the system of the fuel tank planner also includes a data optimization module.

[0019] Preferably, the data optimization module also includes a fuel tank planning optimization submodule and a nomination date optimization submodule.

[0020] Preferably, the system of the fuel tank planner further comprises a notification generating module.

[0021] Preferably, the notification generation module generates any one of a combination of notifications, the combination of notifications comprising: a predicted fuel price and / or fuel index; a fueling contract designation; and an optimized bunkering plan.

[0022] Preferably, the fuel price prediction module receives data from the characteristic data collection submodule and the fuel price data collection submodule.

[0023] Preferably, the fuel price prediction module uses customized machine learning and optimization models for long-term and short-term predictions, and the machine learning and optimization models are selected from: Ensemble Lasso (least absolute shrinkage and selection operator), Random Forest, Distributed Gradient Boosting Library (XGBoost), Support Vector Machine, Trained Structured Prescriptive Empirical Risk Management (SP-ERM) and Trained Truncated Scenario Linear Decision Rule Empirical Risk Management (TSLDR-ERM).

[0024] Preferably, the data optimization module receives data from the data acquisition module and the fuel price prediction module.

[0025] Preferably, the tank planner's system communicates with a server to feed information to the data acquisition module.

[0026] Preferably, the system of the fuel tank planner also includes a human-machine interface system to enable the decision maker to input information into the system of the fuel tank planner, thereby outputting a fuel supply plan.

[0027] In another embodiment, a method for minimizing the refueling cost of a ship is provided, the method comprising: using a system of a data-driven bunker planner located on the ship to predict fuel prices and / or fuel indices. Preferably, the method also comprises: determining the optimal time, quantity and location for refueling, and fulfilling one or more refueling terms contracts based on short-term, long-term or a combination of short-term and long-term predicted fuel prices and / or fuel indices. Preferably, the step of predicting fuel prices and / or fuel indices uses customized machine learning and optimization models for long-term and short-term predictions, and the machine learning and optimization models are selected from: integrated lasso algorithm, random forest, distributed gradient boosting library, support vector machine, trained structured normative empirical risk management (SP-ERM) and trained truncated scenario linear decision rule empirical risk management (TSLDR-ERM).

[0028] One skilled in the art will readily appreciate that the present invention is well adapted to carry out the objects and obtain the objects and advantages mentioned above, as well as those inherent therein.The embodiments described herein are not intended to limit the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To facilitate understanding of the invention, there are shown in the accompanying drawings preferred embodiments, and when considered in conjunction with the following description, the invention, its construction and operation, and its numerous advantages will be readily understood and appreciated.

[0030] Figure 1 The overall infrastructure upon which the systems and methods of the present invention are implemented is shown.

[0031] Figure 2 A detailed block diagram of the system of the fuel tank planner within the computing system in its preferred embodiment is shown.

[0032] Figure 3 An example of information organized within a bunkering terms contract data collection submodule that is part of the data collection module of a system of a bunker planner is shown.

[0033] Figure 4 An example of information organized within a feature data collection submodule as part of a data collection module is shown.

[0034] Figure 5 An example of information organized within a fuel price index data collection submodule as part of a data collection module is shown.

[0035] Figure 6 An example of information organized within a route data collection submodule as part of a data collection module is shown.

[0036] Figure 7 An example of information organized within a fuel consumption data collection submodule as part of a data collection module is shown.

[0037] Figure 8 An example of information organized within a vessel status data collection submodule as part of a data collection module is shown.

[0038] Fig. 9 An example of information organized within a port data collection submodule as part of a data collection module is shown.

[0039] Fig.10 is a flow chart showing the flow of steps performed by a bunker planning optimization submodule that is part of the optimization module of the system as a bunker planner.

[0040] Fig.11 is a flow chart showing the flow of steps performed by a specified date optimization sub-module as part of the optimization module.

[0041] Fig.12 An example of a first output of the notification generation module is shown as a predicted fuel price and / or fuel index.

[0042] Fig.13 An example of a second output of the notification generation module as a refueling contract specified date is shown.

[0043] Fig.14An example of a third output of the notification generation module as a refueling plan is shown. DETAILED DESCRIPTION

[0044] The present invention relates to a system of a data-driven bunker planner for optimizing the bunkering costs of a marine vessel. The present invention can also be presented in a number of different embodiments with common elements. According to the concept of the present invention, a data-driven bunker system is located on a ship, which enables the prediction of fuel prices and / or fuel indices when the ship is traveling along a planned route, and determines the bunkering plan within the optimization of data related to bunkering restrictions, fuel consumption and bunkering facilities at the port of call.

[0045] It should be noted that in the context of the present invention, the term "vessel" may refer to a ship, a container ship, a cargo ship, a bulk carrier or the like.

[0046] It should be noted that in the context of the present invention, the term "port" may refer to a marine port, a harbour or any place where marine vessels are anchored, for example for loading and unloading of cargo.

[0047] Furthermore, it should be noted that ports with fuel suppliers offering refuelling services may be referred to as "refuelling ports" or "fuelling ports".

[0048] Furthermore, it should be noted that ports which are planned stops for a vessel throughout its route may be referred to as “ports of call”.

[0049] In addition, it should be noted that in the context of the present invention, the term "forecast window" may refer to the number of days of the forecast of future fuel prices or indices relative to the forecast time. The term "short-term" may refer to a forecast window of less than or equal to 5 working days; the term "long-term" may refer to a forecast window of more than 5 working days.

[0050] The present invention will now be described in more detail by way of example with reference to the accompanying drawings.For ease of reference, a common reference numeral or series of numerals will be used throughout the drawings when referring to the same or similar features common to the drawings.

[0051] Figure 1 is a diagram showing the overall infrastructure upon which the present invention is implemented. Specifically, Figure 1 The vessel 1 is shown travelling along a pre-planned route (drawn in bold) after departing from a starting port A to reach a destination port Z. The vessel 1 has a computing system 10 which should assist a decision maker of the vessel 1 (e.g., the captain of the vessel) during the journey of the vessel 1 along the pre-planned route.

[0052] Reference Figure 1, along the pre-planned route between the starting port A and the ending port Z, there are multiple ports in between. Some of these ports are considered as ports of call along the pre-planned route, such as Port D, Port E, and Port F. In addition, there may be other ports that provide refueling services near the above-mentioned ports of call D, E, and F, namely Port P, Port Q, and Port R.

[0053] refer to Figure 1 Based on the example itinerary of the pre-planned route, the ship 1 will stop at ports D, E, and F. However, it should be noted that not all ports D, E, and F can provide refueling operations. For example, port D may be a port of call where the ship must stop to unload cargo and refuel, port E may be a port of call where the ship 1 must stop to refuel, and port F may be a port of call where the ship 1 must stop to unload cargo but cannot refuel because port F does not provide refueling services.

[0054] Based on the above considerations, prior to the journey of vessel 1, the decision maker may enter into a bunkering terms contract with the bunker supplier at port D. In a maritime environment, a bunkering terms contract may also be referred to as a bunkering contract. A bunkering terms contract is a contract whereby the decision maker agrees to enter into a contractual obligation with the bunker supplier to purchase fuel from or sell fuel to the bunker supplier at an agreed price. In addition, the bunkering terms contract may have a price basis, which may be a named date (DON) or a delivered date (DOD). For a bunkering terms contract based on a named date (DON), the unit price is determined by the fuel price on the date the bunker fuel is specified (i.e., the date of ordering). For example, the decision maker is made to place an order 5 days before the actual delivery time; therefore, the decision maker needs to predict the fuel price or index for the next 5 days. For a bunkering terms contract based on a delivered date (DOD), the unit price is equal to the fuel price on the date the bunker fuel is delivered relative to the vessel.

[0055] refer to Figure 1 When the ship 1 travels along the pre-planned route, its computing system 10 should obtain the status information of the ship 1 and the information from the outside world. For example, it should establish a multi-channel communication link with the server 2 through the satellite 31 by a known method. The server 2 is preferably located in a stationary position. Since the ship 1 sails on the high seas, the server 2 may be on land or on an anchored buoy.

[0056] In the first use case, the server 2 can act as a data aggregator for the vessel 1 by aggregating information from the outside world. The server 2 can communicate with other servers 3 by wire or wireless means to aggregate data such as fuel prices, fuel index, port information, etc. The server 2 can then transmit the aggregated data (summarized data) back to the vessel 1 via a satellite 31. The computing system 10 of the vessel 1 should then process the status information of the vessel 1 and the aggregated data from the outside world to predict the fuel price and / or fuel index and optimize the refueling cost when the vessel 1 travels along the pre-planned route.

[0057] In a second use case, server 2 may act as a communication intermediary with other servers 3 so that decision makers at vessel 1 may communicate with a fuel supplier to execute a purchase order or a sale order for fuel in accordance with a bunkering terms contract.

[0058] In a third use case, the server 3 can act as a cloud database to store information related to the vessel 1. This may include the vessel's itinerary, current position, parameters, crew data, etc.

[0059] Furthermore, the vessel 1 may include a communication link with a weather satellite 32 to receive information on weather conditions of the ocean along its route.

[0060] Based on the status information of the vessel 1 and the aggregated data from the outside, the computing system 10 can determine the best route. The best route may be the same as the pre-planned route, or it may be an alternative route that takes into account at least the itinerary of the vessel 1, the agreed bunkering terms contract, or the predicted fuel price and / or fuel index at the bunkering port.

[0061] According to the example itinerary, vessel 1 stops at ports D and F to unload cargo and refuels at port D, thereby fulfilling the refueling clause contract. This means that there is a certain degree of freedom in stopping at port E for refueling. Thus, computing system 10 can determine if Figure 1 An example optimal route is shown (position drawn with a dotted line). In the example optimal route, the route to be taken by vessel 1 has been changed from Port D → Port E → Port F to Port D → Port Q → Port F.

[0062] To determine the example optimal route, the computing system 10 must consider a number of factors. A first factor may be the estimated time of arrival (TOA) from the current location of the vessel 1 to the two ports. A second factor may be whether the predicted fuel price at port E (within the estimated TOA range of the vessel) is more expensive than the predicted fuel price at port Q (within the estimated TOA range of the vessel T). In this example optimal route, the computing system 10 may have determined that the estimated TOA of the vessel at the two ports is approximately (more or less) the same, and that the predicted fuel price at port E is more expensive than the predicted fuel price at port Q. Therefore, port E is dropped from the itinerary in the example optimal route and replaced with port Q. Therefore, the computing system 10 will determine a route that optimizes the fuel cost of vessel 1.

[0063] In order to further optimize the fuel costs of the vessel 1, the computing system 10 will check the agreed fuel terms contracts related to the refueling of the vessel 1, for example, every day. Preferably, on each day, the computing system 10 determines whether it is worth fulfilling the fuel terms contract. This is achieved by comparing the predicted fuel price and / or fuel index with the fuel price or fuel index of the day. After checking the fuel terms contract, if the computing system 10 determines that fulfilling the fuel terms contract on the current date will bring profit to the decision maker or the owner of the vessel 1, the computing system 10 will specify the fuel terms contract to be fulfilled. The decision maker can be informed of this through the human-machine interface (HMI) 18 of the vessel 1. According to the example itinerary, this is preferably completed every day before the vessel 1 arrives at port D.

[0064] It should be noted that the computing system 10 of the vessel 1 is a system of systems, wherein it also includes a combination of one or more mechanical, electrical or electronic systems. Each of these systems is substantially interconnected (connected) to allow communication between them. In addition, each of these systems has its own hardware and software components. Preferably, the computing system 10 also includes at least a system 11 of a data-driven fuel tank planner.

[0065] In addition, the computing system 10 may also include a fuel management system, a propulsion system, a ballast system, a cargo tank management system, a navigation system, a power system, a communication system, a data storage system, a positioning system, and a human-machine interface (HMI) system 18. These above-mentioned systems may be implemented in a well-known manner on the vessel 1, and therefore, further explanation of these above-mentioned systems will be minimal.

[0066] Figure 2A detailed block diagram of a system 11 of a data driven bunker planner within a computing system 10 in its preferred embodiment is shown. As shown, the system 11 of a data driven bunker planner is a system within the computing system 10. The computing system 10 may be a conventional computer. The system 11 of a data driven bunker planner may include a processor 11a running application software 11b and at least one storage unit 11c. The system 11 of a data driven bunker planner is a system within the computing system 10 that predicts fuel prices and / or fuel indices and optimizes fueling costs as the vessel 1 travels along its route. To this end, it is essentially interfaced with all other systems of the vessel 1.

[0067] It should be noted that the system 11 of the data-driven fuel tank planner can be implemented in other ways. Since the system 11 of the data-driven fuel tank planner minimally requires only a processor 11a and a memory unit 11c, the system 11 of the data-driven fuel tank planner can also be implemented with or within other systems by sharing hardware components in which the application software 11b can be run. For example, the system 11 of the data-driven fuel tank planner shares the same hardware components with the human-machine interface (HMI) system 18.

[0068] In the system 11 of the data-driven fuel tank planner, the processor 11a performs the scheduling and execution of software instructions or computer logic instructions according to the instructions of the application software 11b running thereon. Preferably, the system 11 of the data-driven fuel tank planner is a conventional processor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or a combination thereof. Regarding the memory unit 11c, it is preferably coupled to the processor 11a. The storage unit 11c can store program files related to the application software 11b, or temporarily store processing data during the operation of the application software 11b.

[0069] Figure 2 It is also shown that the processor 11a operates one or more modules; these modules are software modules of the application software 11b operating or running on the processor 11a, which are used to calculate algorithms and generate control instructions related to fuel supply cost optimization. These modules include a data acquisition module 111, a fuel price prediction module 112, a data optimization module 113, and a notification generation module 114. It should be noted that the modules presented do not need to be software implementations, and can be hardware implementations connected to the processor 11a, or they can be their own independent computer systems. Auxiliary modules can be included to provide auxiliary ports for the above modules.

[0070] In one embodiment, the data acquisition module 111 is used to obtain the status information and aggregated data of the ship 1 from the outside world. Preferably, it also includes a refueling terms contract data acquisition submodule 1111, a feature data acquisition submodule 1112, a fuel price data acquisition submodule 1113, a route data acquisition submodule 1114, a fuel consumption data acquisition submodule 1115, a ship status data acquisition submodule 1116 and a port data acquisition submodule 1117.

[0071] Regarding the refueling terms contract data collection submodule 1111, it is preferred that one or more refueling terms contracts agreed upon between the decision maker and the fuel supplier are obtained before or during the journey. Preferably, the refueling terms contract data is entered by the decision maker through the HMI system 18 of the ship 1. Alternatively, the refueling terms contract data is stored in the server 2 or the ship's data storage system and retrieved therefrom. The refueling terms contract data includes:

[0072] (i) a refuelling or bunkering port for which a bunkering contract has been concluded;

[0073] (ii) the respective fuel suppliers with whom bunkering terms contracts have been concluded at those bunkering ports;

[0074] (iii) an agreed minimum fuel purchase quantity;

[0075] (iv) the agreed maximum fuel purchase quantity;

[0076] (v) fixed service costs imposed by suppliers;

[0077] (vi) the agreed fuel index to be followed;

[0078] (vii) additional costs of the bunkering contract (insurance, surcharges, premiums);

[0079] (viii) The agreed price basis of the bunkering contract, which may be a designated date (DON) or a delivery date (DOD);

[0080] (ix) the estimated number of days the bunkering contract will be in effect until the vessel is expected to arrive at the port; and

[0081] (x) the estimated number of days before the vessel is expected to arrive at the port until the bunkering contract is terminated.

[0082] These data can be organized by the refueling terms contract data collection submodule 1111 into Figure 3 The form shown.

[0083] Regarding the feature data collection submodule 1112, it preferably obtains and / or derives one or more feature item data from one or more data sets, which are considered to have a significant impact on future fuel prices and / or fuel indices. These data sets are preferably aggregated from the outside by the server 2. In one embodiment, the feature data collection submodule 1112 uses known statistical techniques to derive feature item data from the data sets. In one embodiment, the feature data includes:

[0084] (i) the date on which the characteristic item data is or was calculated;

[0085] (ii) first feature item data, which may be a numerical weight corresponding to the impact of West Texas Intermediate (WTI) crude oil prices on fuel prices;

[0086] (iii) second feature item data, which may be a numerical weight corresponding to the impact of the New York Stock Exchange (NYSE) Composite Index on fuel prices;

[0087] (iv) a third feature item data, which may be a numerical weight corresponding to the impact of West Texas Intermediate (WTI) crude oil prices on fuel prices over the previous three days;

[0088] (v) a fourth feature item data, which may be a numerical weight corresponding to the impact of the New York Stock Exchange (NYSE) Composite Index on fuel prices in the previous three days;

[0089] These data can be organized by the feature data collection submodule 1112 into Figure 4 The form shown.

[0090] It should be noted that some feature item data may be considered as "long prediction window" data, while some feature item data may be considered as "short prediction window" data. For example, the first feature item data and the second feature item data are considered as "long prediction window" data because there is no specified range in the data sets from which these data are derived. Among them, the third feature item data and the fourth feature item data can be considered as "short prediction window" data because the data sets from which these data are derived have a specified range in a short period of time (in this case, the first three days).

[0091] In addition, the number of feature item data is not limited to the above. It can be any number that provides an indicator of the impact of the data set on future fuel prices and / or fuel indices. This may include the impact of the fuel refinery capacity of the warehouse on future fuel prices and / or fuel indices, the impact of the fuel refinery output of the warehouse on future fuel prices and / or fuel indices, etc.

[0092] Regarding the fuel price data collection submodule 1113, it preferably obtains one or more fuel prices and / or fuel indices. These fuel prices and / or fuel indices are preferably aggregated from the outside by the server 2. If necessary, historical data of fuel prices and / or fuel indices can also be obtained. For example, the fuel price data includes:

[0093] (i) the date of the fuel price or fuel index;

[0094] (ii) a first fuel index, which may be a value of the Bunker World Index (BWI);

[0095] (iii) a second fuel index, which may be a value of the Bunker World 380 Index (BW380);

[0096] (iv) A third fuel index, which may be a value of the average bunker fuel price in the Asia Pacific (APAC) region.

[0097] These data can be organized by the fuel price data collection submodule 1113 into Figure 5 In addition, the number of fuel prices and / or fuel indices may not be limited to the above numbers, and may be any number thereof.

[0098] Regarding the route data collection submodule 1114, it is preferred that the route information of the ship 1 is obtained based on the ship's itinerary. For example, the route information includes the starting port, the end (destination) port, the port of call for loading or unloading cargo, and the port of call for refueling. In addition, the arrival date and departure date of the ship 1 are also included. This data can be entered by the decision maker into the HMI18 interface of the ship 1, or obtained from the server 2 or the ship's data storage system. This data can be organized by the route data collection submodule 1114 into Figure 6 The form shown.

[0099] Regarding the fuel consumption data acquisition submodule 1115, it is preferably based on the information obtained from other systems in the computing system 10 to obtain the fuel consumption data of the ship 1. For example, the fuel consumption data includes:

[0100] (i) port-to-port details;

[0101] (ii) The estimated distance between one port and another.

[0102] (iii) the estimated speed of the vessel when travelling between ports.

[0103] (iv) the estimated fuel consumption of the vessel when travelling between one port and another.

[0104] These data can be organized by the fuel consumption data collection submodule 1115 into Figure 7 The form shown.

[0105] Regarding the ship status data acquisition submodule 1116, it preferably obtains the current status of the ship at any given time during the ship's voyage based on information obtained from other systems within the computing system 10. For example, the ship status data includes:

[0106] (i) the name of the ship;

[0107] (ii) the International Maritime Organization (IMO) number of the vessel.

[0108] (iii) the intended service of the vessel;

[0109] (iv) timestamp of the status update;

[0110] (v) the planned speed of the vessel;

[0111] (vi) the actual speed of the vessel;

[0112] (vii) forecast weather conditions along the current route, which may also include wind speed, sea temperature, etc.;

[0113] (viii) the quantity of cargo in the vessel’s cargo holds.

[0114] (ix) the amount of fuel remaining on board (ROB);

[0115] (x) the amount of fuel consumed;

[0116] (xi) Estimated distance from current location to another port.

[0117] These data can be organized by the fuel consumption data collection submodule 1116 into Figure 8 The form shown.

[0118] Regarding the port data collection submodule 1117, it preferably obtains data of ports providing refueling services located along or near the pre-planned route. These data are preferably aggregated from the outside by the server 2. For example, the port data includes:

[0119] (i) the name of the terminal at the port;

[0120] (ii) coordinates of the port;

[0121] (iii) the type of fuel available at the port, which may include very low sulphur fuel oil (VLSFO), heavy fuel oil (HFO) and ultra low sulphur fuel oil (ULSFO);

[0122] (iv) The waterline of the port.

[0123] These data can be organized by the port data collection submodule 1117 into Fig. 9 The form shown.

[0124] In one embodiment, the fuel price prediction module 112 is used to predict fuel prices based on the data provided by the feature data acquisition submodule 1112 and the fuel price acquisition submodule 1113. Preferably, it also includes a long-term price prediction submodule 1121 and a short-term price prediction submodule 1122. In one embodiment, these fuel price prediction submodules 1121, 1122 use various customized machine learning and optimization models to perform long-term prediction and short-term prediction, and the machine learning and optimization models are selected from: integrated lasso algorithm, random forest, distributed gradient boosting library, support vector machine, trained structural normative empirical risk management (SP-ERM) and trained truncated scenario linear decision rule empirical risk management (TSLDR-ERM). According to the prediction window (or the number of days of future fuel prices or indices relative to the prediction time), the best performing model is selected and integrated into the prediction submodules 1121, 1122. Preferably, this prediction method includes: (i) long-term decision; and (ii) short-term decision. Long-term decisions are based on long-term forecasts of fuel prices and / or fuel indices, for example, when the forecast window exceeds 5 days, the optimal combination of contracts is determined for a given port contract combination; short-term decisions are based on short-term forecasts, for example, when the forecast window is shorter than or equal to 5 days, the optimal time for ordering is determined based on the terms and contracts specified in the refueling terms contract data in submodule 1111.

[0125] Regarding the long-term prediction submodule 1121, it preferably uses feature data that is considered as data within the long prediction window period, such as the first feature item data and the second feature item data mentioned above, to generate a value representing the fuel price and / or fuel index within the long prediction window. In another embodiment, the long-term prediction may refer to a prediction window longer than 5 days. In addition, the planning period for long-term decisions may be basically one month, which means that the decision maker needs to make a refueling decision (when, where and how much to buy) for the next month.

[0126] Regarding the short-term prediction submodule 1122, it preferably uses feature data that is considered as data on a short prediction window, such as the third feature item data and the fourth feature item data mentioned above, to generate a numerical value representing the fuel price and / or fuel index in the short prediction window. In another embodiment, a short-term prediction may refer to a prediction window shorter than or equal to 5 days, which means that the decision needs to predict the fuel price within the next 5 days and make the refueling order 5 days before the actual delivery date. Short-term predictions require higher accuracy but are relatively easy to decide; long-term predictions are more challenging and may use various customized machine learning and optimization models to predict longer window periods.

[0127] In one embodiment, the data optimization module 113 is used to optimize the refueling cost when the ship 1 travels along its itinerary. Preferably, it also includes a fuel tank planning optimization submodule 1131 and a designated date optimization submodule 1132.

[0128] More specifically, the fuel tank planning optimization submodule 1131 is used to determine the time, quantity and location of fuel replenishment to minimize the fuel replenishment cost. Among them, the designated date optimization submodule 1132 will specify the best time to place an order after receiving the refueling terms contract. With these two sets of information, the decision maker can operate the ship 1 along the planned route while minimizing the fuel replenishment cost.

[0129] Fig.10 A series of operation processes performed by the fuel tank planning optimization submodule 1131 are shown. The fuel tank planning optimization submodule 1131 will perform a series of steps to determine the time, quantity and location of fuel replenishment, wherein the fuel replenishment cost can be minimized using the information from the data acquisition module 111 and the predicted long-term fuel price and / or fuel index from the long-term prediction submodule 1121 of the fuel price prediction module 112.

[0130] First, in step A1, the decision maker of the ship 1 interacts with the HMI system 18 of the ship 1 to define a planning horizon for a future period of time for the bunker planning optimization submodule 1131. The planning horizon can be a customizable time frame. For example, a schedule with a time of 90 days from the current location, a schedule of 180 days after the ship 1 leaves the port, a schedule in which the ship stops at least two future ports during the itinerary, etc.

[0131] Next, in step A2, the fuel tank planning optimization submodule 1131 simulates the future actions of the ship 1 within the defined planning scope. For example, it simulates the time for the ship 1 to arrive at the port based on the route data obtained from the route data collection submodule 1114 and the current state of the ship 1 obtained from the ship state data collection submodule 1116.

[0132] Next, in step A3, the bunker planning optimization submodule 1131 calculates the fuel price per unit for each bunkering port visit. This is done based on the simulation information from step A2, the bunkering terms contract information, and the predicted long-term fuel price and / or fuel index.

[0133] Next, in step A4, the fuel tank planning optimization submodule 1131 obtains information about the current remaining on board (ROB) fuel amount of the ship 1 from the ship status data collection submodule 1116.

[0134] Next, in step A5, the fuel planning optimization submodule 1131 identifies one or more fueling constraints that may affect future ROB fuel volumes, particularly during the defined planning horizon. For example, these fueling constraints may include:

[0135] (i) The capacity of the fuel tank;

[0136] (ii) the usage rate of different types of fuel between fuel tanks;

[0137] (iii) future weather conditions along the route;

[0138] (iv) the minimum and maximum fuel purchase quantities specified in the bunkering terms contract;

[0139] (v) speed of vessel 1;

[0140] (vi) the cargo remaining in the cargo hold of the vessel;

[0141] (vii) Electricity consumption of vessel 1.

[0142] Next, in step A6, the tank planning optimization submodule 1131 attempts to minimize the refueling costs within the defined planning horizon based on these refueling constraints.

[0143] Next, in the final step A7, the fuel tank planning optimization submodule 1131 generates a fuel supply plan within the defined planning scope and refueling restrictions.

[0144] refer to Figure 1 As described above, the optimal route can be generated through the above series of steps performed by the fuel tank planning optimization submodule 1131.

[0145] Fig.11 The B series of operation processes performed by the designated date optimization submodule 1132 are shown. The designated date optimization submodule 1132 will perform a series of steps to enable the decision maker to fulfill its contractual obligations at the best time based on the data from the refueling terms contract acquisition submodule 1111 and the predicted short-term fuel price and / or fuel index from the short-term price prediction submodule 1122 of the fuel price prediction module 112.

[0146] In step B1 , the designated date optimization submodule 1132 generates an ordered list of ports that have signed bunkering terms contracts, sorted by the scheduled arrival time of vessel 1 .

[0147] Next, in step B2, the designated date optimization submodule 1132 indexes each port in the list as a task i that needs to be performed. i is defined as an ordered sequence of 1, 2, ... N, where N is the number of ports in the list.

[0148] Next, in step B3, the designated date optimization submodule 1132 selects a task to be completed.

[0149] The next step of step B3 is step B4. Step B4 is a decision step, in which it should be determined whether all tasks have been marked as "completed". If not, step B4 will proceed to step B5. If this is the case, step B4 proceeds to step B15.

[0150] In step B5, the designated date optimization submodule 1132 retrieves further details of the mission, such as the corresponding bunkering terms contract of the port and the estimated arrival time of the vessel 1 at the port.

[0151] The step following step B5 is step B6. Step B6 is another decision step where it is determined whether the bunkering terms contract for the mission has an agreed price basis, namely "DON". If this is the case, step B6 proceeds to step B7. If not, step B6 will proceed to step B16.

[0152] Step B7 is another decision step, in which it should be determined whether the refueling terms contract based on the price basis of the "specified date" has entered the specified period based on the current date. If this is the case, step B7 proceeds to step B8. If not, step B7 will proceed to step B17.

[0153] Step B8 is another decision step, in which it should be determined whether the refueling terms contract based on the price basis of the "specified date" has reached the end date of the specified period based on the current date. If not, step B8 will jump to step B9. If this is the case, step B8 proceeds to step B18.

[0154] In step B9, the specified date optimization submodule 1132 retrieves the predicted short-term fuel price and / or fuel index and fits it to the current window of the specified period. The current window is defined as the duration between the current date and the end date of the specified period.

[0155] Next, in step B10 , the designated date optimization submodule 1132 compares the predicted short-term fuel price and / or fuel index with the fuel price and / or fuel index of the current date.

[0156] The step following step B10 is step B11. Step B11 is another decision step in which it is determined whether the fuel price and / or fuel index on the current date is lower than the predicted short-term fuel price and / or fuel index. If not, step B11 proceeds to step B12. If this is the case, step B11 proceeds to step B13.

[0157] In step B12, since it is determined that the fuel price and / or fuel index on the current date is more expensive than the predicted short-term fuel price and / or fuel index, the task will be marked as "uncompleted" and retained as a task to be reselected later.

[0158] In step B13, since the fuel price and / or fuel index on the current date is determined to be cheaper than the short-term fuel price and / or fuel index predicted by the price prediction module 112, the current date is considered to be the best time for the refueling terms contract to be fulfilled. Therefore, the task is assigned by the assigned date optimization submodule 1132 so as to fulfill the refueling terms contract on the current date.

[0159] The step after step B13 is step B14. In B14, the task will be marked as "completed" and its corresponding index will be ignored by the specified date optimization submodule 1132 from now on.

[0160] Steps B16, B17 and B18 will now be described.

[0161] In step B16, since the refueling terms contract for the determined task has no agreed upon price basis, namely "DON", the DON optimization submodule 1132 interprets it as completely ignoring the task.

[0162] Therefore, following step B16 is step B14, where the task will be marked as "completed" and its corresponding index will be ignored from here on by the designated date optimization sub-module 1132.

[0163] In step B17, since it is determined that the refueling terms contract based on the price basis of the "specified date" has not yet entered the specified period based on the current date, the specified date optimization submodule 1132 interprets it as temporarily ignoring the task.

[0164] Thus, following step B16 is step B12, where the task will be marked as "not completed" and retained as a task to be reselected later.

[0165] In step B18, since it is determined that the refueling terms contract based on the price basis of the "designated date" has reached the end date of the designated period based on the current date, the designated date optimization submodule 1132 interprets that an immediate designated task is required.

[0166] Therefore, following step B18 is step B13, in which the task is specified by the specified date optimization submodule 1132 so as to fulfill the terms of the contract on the current date. Again, following step B13 is step B14. In B14, the task will be marked as "completed" and its corresponding index will be ignored by the specified date optimization submodule 1132 from now on.

[0167] Step B12 or B14 is followed by step B3. This means that after marking a task as "not completed" or "completed", the designated date optimization submodule 1132 selects a task again as a subsequent task to be completed. Preferably, the designated date optimization submodule 1132 selects the subsequent task in a sequential manner based on the task index. This can be achieved by incrementing the task index i by 1, so that the index of the next task should be i+1. At this point, step B3 advances to step B4 again, and the entire process flow begins to perform the following tasks.

[0168] Step B15 will now be described, which can be considered as the last step in the B series operation flow. In step B15, since it is determined that all tasks have been marked as "completed", the designated date optimization submodule 1132 interprets that all tasks have been completed. At this point, the B series operation flow ends. The designated date optimization submodule 1132 can then notify the decision maker through the HMI system 18 of the vessel 1 that all tasks have been completed.

[0169] The notification generation module 114 is used to generate one or more notifications, which are presented to decision makers through the HMI system 18 of the vessel 1. For example, the notification generation module 114 should output at least three different notifications, including: (i) predicted fuel prices and / or fuel indexes; (ii) bunkering terms contract specifications; and (iii) fuel supply planning for the remaining trip.

[0170] The predicted fuel price and / or index is a first output of the notification generation module 114 and can be organized by the vessel's HMI system 18 to Fig.12 In one embodiment, the predicted fuel price includes:

[0171] (i) How many days are left until the current date;

[0172] (ii) one or more predicted fuel prices and / or fuel indices.

[0173] For example, regarding how many days are left until the current date, the number of days may vary depending on the configuration of the application software 11b. As for the fuel price and / or fuel index, they may come from one or both of the long-term prediction submodule 1121 or the short-term prediction submodule 1122 of the price prediction module 112. Preferably, the fuel price and / or index is predicted by removing and combining the long-term and short-term predicted fuel prices and / or indices obtained from the long-term prediction submodule 1121 and the short-term prediction submodule 1122. For example, the adopted fuel price and / or fuel index may be the middle value of the long-term and short-term predicted fuel prices and / or indices.

[0174] For example, for the case where the number of days away is <40 days, the predicted short-term fuel price and / or index from the short-term prediction submodule 1122 will be used. Wherein, for the case where the number of days away is >40 days, the predicted long-term fuel price and / or index from the long-term prediction submodule 1121 will be used.

[0175] The bunkering contract designation is a second output of the notification generation module 114 and may be organized by the vessel's HMI system 18 to Fig.13 For example, the bunkering clause contract specifies:

[0176] (i) Mission index for the refueling contract;

[0177] (ii) the estimated time of arrival of the vehicle at the port where the bunkering terms have been agreed;

[0178] (iii) the fuel price and / or fuel index for the current date;

[0179] (iv) a fuel price and / or fuel index predicted by the descending time of arrival (TOA) of vessel 1;

[0180] (v) instructions should specify the indicators for fulfilling the bunkering terms contract;

[0181] (vi) Indicators that bunkering clause contracts should continue as a mandate.

[0182] For example, the predicted fuel price and / or fuel index in the incoming TOA of vessel 1 comes from the short-term prediction submodule 1122 of the price prediction module 112 .

[0183] The refueling plan is a third output of the notification generation module 114 and can be organized by the vessel's HMI system 18 to Fig.14 For example, the fuel supply plan is an ordered list of the status of each tank in the ship 1, preferably including at least:

[0184] (i) the name of the ship;

[0185] (ii) the intended service of the vessel;

[0186] (iii) itinerary serial number;

[0187] (iv) the available fuel tanks on board the vessel, each of which is assigned an index number;

[0188] (v) The type of fuel acceptable to the tank;

[0189] (vi) Maximum capacity of the fuel tank;

[0190] (vii) the preferred maximum fill level of the fuel tank;

[0191] (viii) the preferred maximum fill volume of the fuel tank (expressed as a percentage);

[0192] (ix) Volume of waste oil in the tank;

[0193] (x) the port of call at which Vessel 1 will call;

[0194] (xi) Estimate the time of arrival (TOA) of vessel 1 at each port;

[0195] (xii) fuel consumed by vessel 1 upon arrival at the port;

[0196] (xiii) the fuel remaining on board (ROB) of Vessel 1 when it arrives at the port;

[0197] (xiv) fuel that needs to be restocked or replenished in tanks;

[0198] (xv) The amount of fuel remaining on board (ROB) when the vessel leaves port.

[0199] Thus, the details related to the computing system 10 and the system 11 of the data driven bunker planner of the present invention have been fully explained. In addition to the advantages described above, the present invention provides an improved management of ship operations in a marine environment. In addition, the present invention provides decision makers with further flexibility to manage fuel as a derivative, so that they can effectively adopt risk management strategies such as fuel price hedging.

[0200] The present disclosure includes the appended claims and the foregoing description. Although the present invention has been described with a certain degree of particularity in its preferred form, it should be understood that the disclosure of the preferred form is by way of example only and that many changes may be made to the structural details and the combination and arrangement of parts without departing from the scope of the present invention.

Claims

1. A system for a data-driven tank planner, wherein: The data driven tank planner system comprises: A system (11) of a bunker planner, the system (11) of the bunker planner being operable in a computing system (10) located in a vessel (1); The system (11) is capable of operating the bunker planner to predict fuel prices and / or fuel indices, determine the best time, quantity and location for refueling, and fulfill one or more refueling terms contracts based on the predicted fuel prices and / or fuel indices.

2. The system according to claim 1, wherein: The fuel tank planner system (11) also includes a data acquisition module (111).

3. The system according to claim 2, wherein: The data acquisition module (111) further comprises: Refueling terms contract data collection submodule (1111); Feature data collection submodule (1112); Fuel price data collection submodule (1113); Route data collection submodule (1114); Fuel consumption data collection submodule (1115); Ship status data collection submodule (1116); and Port data collection submodule (1117).

4. The system according to claim 2 or 3, wherein: The fuel tank planner system (11) also includes a fuel price prediction module (112).

5. The system according to claim 4, wherein: The fuel price prediction module (112) further includes: Long-term price prediction submodule (1121); and Short-term price prediction submodule (1122).

6. The system according to claim 4 or 5, wherein: The fuel tank planner system (11) also includes a data optimization module (113).

7. The system according to claim 6, wherein: The data optimization module (113) further includes: Fuel tank planning optimization submodule (1131); and The date optimization submodule (1132) is specified.

8. The system according to claim 6 or 7, wherein: The tank planner system (11) further comprises a notification generating module (114).

9. The system according to claim 8, wherein: The notification generating module (114) generates any one of a combination of notifications, the combination of notifications comprising: forecasted fuel prices and / or fuel indices; The bunkering terms contract specifies; and Refueling planning.

10. The system according to any one of claims 4 to 9, wherein: The fuel price prediction module (112) receives data from the characteristic data collection submodule (1112) and the fuel price data collection submodule (1113).

11. A system according to any one of claims 4 to 10, wherein: The fuel price prediction module (112) uses various customized machine learning and optimization models to perform long-term and short-term predictions, and the machine learning and optimization models are selected from: integrated lasso algorithm, random forest, distributed gradient boosting library, support vector machine, trained structured normative empirical risk management (SP-ERM) and trained truncated scenario linear decision rule empirical risk management (TSLDR-ERM).

12. A system according to any one of claims 6 to 11, wherein: The data optimization module (113) receives data from the data collection module (111) and the fuel price prediction module (112).

13. A system according to any one of claims 3 to 12, wherein: The system (11) of the fuel tank planner communicates with the server (2) to obtain data for the data acquisition submodules (1111-1117).

14. A system according to any one of claims 9 to 13, wherein: The fuel tank planner system (11) also includes a human-machine interface system (18) to enable a decision maker to input information into the fuel tank planner system (11) and enable the fuel tank planner system (11) to output the fuel supply plan.

15. A method for optimizing the bunkering costs of a vessel (1), wherein: The method comprises using a system (11) of a bunker planner located on board the vessel (1) to predict a fuel price and / or a fuel index.

16. The method according to claim 15, wherein: The method also includes determining the best time to place orders in the short term, or determining the best combination of contracts in the long term under a given fueling terms contract.

17. The method according to claim 15 or 16, wherein: The step of predicting fuel prices and / or fuel indices employs various customized machine learning and optimization models for long-term and short-term predictions, wherein the machine learning and optimization models are selected from: an integrated lasso algorithm, a random forest, a distributed gradient boosting library, a support vector machine, a trained structured prescriptive empirical risk management (SP-ERM), and a trained truncated scenario linear decision rule empirical risk management (TSLDR-ERM).

18. The method according to any one of claims 15 to 17, wherein: The method further comprises outputting a fuel supply plan based on the predicted fuel price and / or fuel index and the result of data optimization related to the planned route, fuel consumption, ship status and port of call data.

Images