System and method for training ai model for predicting performance of wind propulsion device

Abstract

Disclosed is a method for training an artificial intelligence model for predicting a performance of at least one wind propulsion device (202) on a vessel (204). The method comprising configuring a processor (210) for: receiving force measurement data indicative of a force generated by the at least one wind propulsion device; receiving data of at least one operational parameter indicative of an operating condition associated with at least one of: the at least one wind propulsion device, the vessel; and training the artificial intelligence model with the force measurement data and the data of the at least one operational parameter for establishing a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

Description

[0001] SYSTEM AND METHOD FOR. TRAINING Al MODEL FOR PREDICTING PERFORMANCE OF WIND PROPULSION DEVICE

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to methods for training artificial intelligence (Al) models for predicting performance of wind propulsion devices. The present disclosure also relates to systems for training the Al models for predicting performance of wind propulsion devices.

[0004] BACKGROUND

[0005] In recent years, there have been significant developments in maritime industry to utilise wind energy for propulsion of vessels, for example, such as tanker vessels, cargo vessels, passenger vessels, boats, and the like. Wind propulsion devices such as magnus-rotors or aerofoil sails, are increasingly being used to optimize a fuel efficiency and assist conventional propulsion systems (such as vessel engines). Typically, the wind propulsion devices are vertically installed onto the vessels, and generate a lift that acts as a propulsion force for propelling the vessels in water bodies. However, it is crucial to accurately predict performance with different wind directions, wind speed, and the like, to ensure that the wind propulsion devices would work efficiently in diverse marine environments.

[0006] Some existing technologies for predicting performance of the wind propulsion devices are highly unreliable and error-prone. This is because such existing technologies predominantly rely on physics-based models, for example, such as computational fluid dynamics (CFD) simulations, finite element analysis (FEA), wind tunnel experiments, and the like. Such physics-based models simulate a behavior of the wind propulsion devices under various conditions (i.e., how a wind propulsion device will behave in real-world scenarios), by using input parameters such as wind speed, vessel characteristics, and environmental conditions. However, these physics-based models are often able to provide minimally accurate, minimally realistic results, and may also require human intervention for evaluating effects of the input parameters in forecasting the performance of the wind propulsion devices. Moreover, some existing technologies also utilise measurement biases in conjunction with the physics-based models, for predicting the performance of the wind propulsion devices. However, an overall accuracy and reliability of such a prediction is significantly low.

[0007] Therefore, in light of the foregoing discussion, there exists a need to overcome the aforementioned drawbacks.

[0008] SUMMARY

[0009] The present disclosure seeks to provide a method and a system to realistically, accurately, and reliably predict a performance of a wind propulsion device, in an energy-efficient and time-efficient manner. The aim of the present disclosure is achieved by a method and a system for training artificial intelligence (Al) model for predicting a performance of a wind propulsion device, as defined in the appended independent claims to which reference is made to. Advantageous features are set out in the appended dependent claims.

[0010] Throughout the description and claims of this specification, the words "comprise" , "include", "have", and "contain" and variations of these words, for example "comprising" and "comprises" , mean "including but not limited to", and do not exclude other components, items, integers or steps not explicitly disclosed also to be present. Moreover, the singular encompasses the plural unless the context otherwise requires. In particular, where the indefinite article is used, the specification is to be understood as contemplating plurality as well as singularity, unless the context requires otherwise.

[0011] BRIEF DESCRIPTION OF THE DRAWINGS

[0012] FIG. 1 illustrates steps of a method for training an artificial intelligence model for predicting a performance of at least one wind propulsion device on a vessel, in accordance with an embodiment of the present disclosure; and

[0013] FIG. 2 illustrates an exemplary scenario in which a system for training an artificial intelligence model for predicting a performance of at least one wind propulsion device on a vessel, is employed, in accordance with an embodiment of the present disclosure.

[0014] DETAILED DESCRIPTION OF EMBODIMENTS

[0015] The following detailed description illustrates embodiments of the present disclosure and ways in which they can be implemented. Although some modes of carrying out the present disclosure have been disclosed, those skilled in the art would recognize that other embodiments for carrying out or practising the present disclosure are also possible.

[0016] In a first aspect, an embodiment of the present disclosure provides a method for training an artificial intelligence model for predicting a performance of at least one wind propulsion device on a vessel, wherein the method comprises: receiving, by a processor, force measurement data indicative of a force generated by the at least one wind propulsion device; receiving, by a processor, data of at least one operational parameter indicative of an operating condition associated with at least one of: the at least one wind propulsion device, the vessel; and training, by a processor, the artificial intelligence model with the force measurement data and the data of the at least one operational parameter for establishing a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

[0017] In a second aspect, an embodiment of the present disclosure provides a system for training an artificial intelligence model for predicting a performance of at least one wind propulsion device on a vessel, the system comprising: one or more first sensors for determining force measurement data indicative of a force generated by the at least one wind propulsion device; one or more second sensors for determining data of at least one operational parameter indicative of an operating condition associated with at least one of: the at least one wind propulsion device, the vessel; and a processor operatively coupled to the one or more first sensors and the one or more second sensors, wherein the processor is configured to: receive the force measurement data; receive the data of the at least one operational parameter; and train the artificial intelligence model with the force measurement data and the data of the at least one operational parameter for establishing a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

[0018] The present disclosure relates to the aforementioned method and the aforementioned system for realistically, accurately, and reliably predicting the performance of the at least one wind propulsion device, in an energy-efficient and time-efficient manner that is applicable with minimal modifications from vessel to vessel. Herein, the method and the system utilised the Al model to provide accurate performance forecasts for different control inputs within the wind propulsion device control. Moreover, the method and the system facilitates in providing informed decision-making for vessel manoeuvring, supports optimal vessel power management choices, and contributes to accurate route optimization based on the predictive relationship. Beneficially, by utilizing the Al model, the method and the system are susceptible to improve an overall fuel efficiency of the vessel, and to maximize energy efficiency of the at least one wind propulsion device, thereby managing an electrical power (from for example, a fuel source, a battery, a renewable source) available on the vessel judiciously. The aforementioned method and system allow the wind propulsion device to harness wind power to propel the vessel (to move in a desired course on the waterbody), with minimal energy consumption, leading to lower overall fuel consumption and cost savings. Thus, the aforementioned method and system aid in reducing greenhouse gas emissions, contributing to more sustainable and eco-friendly sailing practices. Additionally, the Al model when trained with a comprehensive dataset (as discussed above), ensures accurate predictions for performance of the at least one wind propulsion device under various wind and operating conditions. The aforementioned method and the aforementioned system are simple, robust, fast, reliable, and can be implemented with ease.

[0019] The method enables in training the artificial intelligence (Al) model by providing the force measurement data and the data of at least one operational parameter as an input (namely, training data) to the Al model, wherein upon training, the Al model is used for predicting the performance of the at least one wind propulsion device. It will be appreciated that the Al model may be designed to forecast and assess how effectively the at least one wind propulsion device would operate on the vessel under various conditions, for example, such as an atmospheric pressure, a speed of wind surrounding the vessel, a speed of the vessel, a direction of a flow of the wind, a direction of a movement of the vessel, and so forth.

[0020] Notably, the processor controls an overall operation of the system. Optionally, the processor is communicably coupled to at least the one or more first sensors and the one or more second sensors. Optionally, the processor is implemented as a processor of a computing device. Examples of the computing device include, but are not limited to, a laptop, a desktop, a tablet, a phablet, a personal digital assistant, a workstation, and a console. Alternatively, optionally, the processor is implemented as a cloud server (namely, a remote server) that provides a cloud computing service.

[0021] Throughout the present disclosure, the term "vessel" refers to a watercraft (namely, a ship) that is used for navigation on a water body (for example, such as an ocean, a sea, and the like). Typically, vessels are used for purposes, for example, such as transportation of cargo and passengers, sea exploration, and the like. Moreover, the vessels are propelled using engines or turbines.

[0022] Throughout the present disclosure the term "wind propulsion device" refers to a device that is capable of harnessing wind power to propel the vessel on the water body. Generally, in operation, the at least one wind propulsion device captures a kinetic energy of wind around the vessel, thereby creating aerodynamic or aerostatic forces that push the vessel forward. This is because as the wind flows over the surface of the at least one wind propulsion device, it creates differences in air pressure around the at least one wind propulsion device. Such a pressure differential generates lift or thrust that is transferred to the vessel, propelling it in a direction of prevailing wind. It will be appreciated that the wind propulsion device typically facilitates in reducing fuel consumption of the vessel, reducing emissions from the vessel, and minimising an operating cost of the vessel, for example, by assisting a primary propulsion means of the vessel, such as an engine or a turbine. However, the at least one wind propulsion device may also be used as a sole propulsion means of the vessel. Examples of the at least one wind propulsion device include, but are not limited to, a magnus rotor (such as a Flettner rotor) and an aerofoil sail (such as wing sails). The aerofoil sail could, for example, be a wing sail, a rigid sail, a suction sail (may also be called a turbo sail), a Bermuda sail, and the like.

[0023] Throughout the present disclosure the term "performance" refers to a relative assessment of at least one wind propulsion device, based on the wind power available (because of the airflow) that can be harnessed and the wind power that is actually harnessed and converted into mechanical power to propel the vessel on the water body. For example, the performance of the at least one wind propulsion device may be high when a major portion of the wind power available is harnessed, and the performance of the at least one wind propulsion device may be low when the wind power available is not harnessed effectively. It may be appreciated that the performance of the at least one wind propulsion device is predicted in term of energy-efficiency and time-efficiency of the at least one wind propulsion device.

[0024] Throughout the present disclosure, the term "control input" refers to at least the force measurement data and the data of the at least one operational parameter. Optionally, the control inputs may be various inputs pertaining to operational adjustment of the at least one wind propulsion device and components thereof, namely, one or more windcapturing components and a power transmission mechanism. Optionally, the different control inputs refer to at least one of: rotor torque control, rotor brake control, and so on. The different control inputs are adjusted to maximize the wind power harnessed. For example, the rotor rotation speed may be adjusted, in order to harness more wind power and / or to minimize damage in extreme weather condition (such as highly windy weather, tornado, cyclone and so on). In another example, the yaw (which refers to a housing of the wind capturing component) may be adjusted to align the at least one wind propulsion device with direction of wind, to maximize wind power harnessing. In yet another example, the rotor torque may be controlled for efficient conversion of the wind power to mechanical power.

[0025] Throughout the present disclosure, the term "force measurement data" refers to information pertaining to forces exerted by the wind on the at least one wind propulsion device, when the vessel is in operation (namely, moving in the water body). In this regard, the force measurement data is obtained, for example, from sensors arranged on the at least one wind propulsion device (as discussed hereinbelow). The force measurement data may comprise at least one of: a magnitude of the forces exerted by the wind, a direction of the forces exerted by the wind. The magnitude of the forces exerted by the wind may be values representing a thrust or a resistance produced by the at least one wind propulsion device, in various wind conditions. Herein, the term "wind condition" refers to a state (namely, a speed, a direction, and / or a turbulence) of the wind at a given location and a given time. It will be appreciated that the force measurement data is received (by the processor) in real-time or near- real time.

[0026] Optionally, the force measurement data is determined using one or more first sensors, and wherein the force measurement data comprises at least one of: pressure-based force measurement data, strain-based force measurement data, displacement-based force measurement data.

[0027] The term "first sensor" refers to a device that is capable of detecting and / or measuring a force exerted by the wind on (one or more surfaces of) the at least one wind propulsion device. Optionally, at least some of the one or more first sensors are arranged on the at least one wind propulsion device. It will be appreciated that some first sensors may be arranged on the at least one wind propulsion device, and remaining first sensors may be arranged on other different parts of the vessel. Different types of the one or more first sensors may be employed for measuring different types of forces (for example, such as a pressure-based force, a strain-based force, a displacement-based force). In this regard, the different types of the one or more first sensors could be arranged on different surfaces of and / or at different locations within the at least one wind propulsion device accordingly.

[0028] The term "pressure-based force measurement data" refers to information pertaining to a pressure (namely, a force per unit area) exerted by the wind on a surface of the at least one wind propulsion device. In this regard, a given first sensor is optionally implemented as a pressure sensor. Pressure sensors and their working are well-known in the art. Typically, when the wind flows over surfaces of the at least one wind propulsion device whereat the given first sensor is arranged, the wind creates varying levels of pressure at those surfaces, and such varying levels of the pressure are detected / measured by the given first sensor. Furthermore, the pressure-based force measurement data may be in form of numerical values of air pressure which may, for example be, expressed in units of Pascal (Pa), millimetre of Mercury (mmHg), pounds per square inch (psi), or similar. Examples of the one or more first sensors to be employed for measuring the pressure-based force measurement data may include, but are not limited to, a piezo-resistive pressure sensor, a capacitive pressure sensor, a microelectromechanical systems (MEMS) pressure sensor, and an optical pressure sensor.

[0029] The term "strain-based force measurement data" refers to information pertaining to a deformation or a strain experienced by a surface (namely, a structure or a material) of the at least one wind propulsion device, when a force is exerted by the wind on the at least one wind propulsion device. Said strain could, for example, be a longitudinal strain, a lateral strain, a volumetric strain, a shear strain, or similar. Types of the strain are well- known in the art. It will be appreciated that the one or more first sensors could be employed as strain gauges (namely, strain-based sensors), for measuring the strain-based force measurement data. The strain gauges are typically attached or bonded to a surface of an object where forces are expected, and as the surface deforms under stress, the strain gauge attached to the surface also deforms with it, causing a change in an electrical resistance of the strain gauges. The strain gauges convert said change in the electrical resistance into an electrical signal that is proportional to the (applied) strain.

[0030] For example, when the at least one wind propulsion device is implemented as a soft sail-based propulsion device, and the strain gauge is arranged on a surface of the soft sail-based propulsion device. In such a case, the strain gauge may record a strain of 0.1 percent, which means the surface may have deformed by 0.1 percent when the soft sail-based propulsion device interacts with the wind. Examples of such strain or deformation may include, but are not limited to, misalignment of rotor of rotor-based propulsion device, surface damage (for example, such as dents, cracks) on suction-based propulsion device or hard sail-based propulsion device, wrinkles in sail of soft sail-based propulsion system, and tangles in the cables of kite-based propulsion system.

[0031] The term "displacement-based force measurement data" refers to information obtained by measuring a movement or a change in a position of at least a part of the at least one wind propulsion device under influence of external wind forces. In this regard, a given first sensor may be implemented as a displacement-based sensor, wherein the displacement-based sensor is employed to detect and quantify an extent of a displacement of said part caused due to the external wind forces. It will be appreciated that some specific parts of the at least one wind propulsion device may be identified where force displacements are expected to be significant, and these parts may be served as a baseline that indicates an initial state of the at least one wind propulsion device, prior to occurrence of the force displacements. In other words, the baseline provides a reference point for understanding subsequent force displacements caused by the external wind forces. Therefore, when the displacement-based sensor experiences the external wind forces, it measures subsequent force displacements. It will be appreciated that the displacement-based sensor may continuously monitor the identified part(s) of the at least one wind propulsion device, and generate electrical signal outputs that correspond to a magnitude and a direction of a corresponding force displacement. The displacement-based sensor may be implemented as a linear variable differential transformer (LVDT), an accelerometer, an optical displacement sensor, and the like. The displacement-based force measurement data may also be referred to as dislocation / deflection-based force measurement data.

[0032] In some implementations, the processor is configured to obtain the force measurement data (which refers to information on force exerted by wind on at least one wind propulsion device in different weather conditions) directly from the one or more first sensor, in real-time, and incorporate such data in training the Al model thereby improving efficiency of the Al model for accurate forecasting. In other implementations, the processor is configured to obtain the force measurement data from a data repository whereat the force measurement data is stored, the processor being communicably coupled to the data repository. A technical benefit of determining the force measurement data is that the force measurement data facilitates in improving adaptability, data richness, and accuracy in establishing the predictive relationship between the performance of the at least one wind propulsion device with the different control inputs.

[0033] Throughout the present disclosure, the term "operational parameter" refers to a variable representing an operating condition of the at least one of: at least one wind propulsion device, the vessel. It will be appreciated that the operating conditions could vary depending on factors, for example, such as a wind speed, a sea state, a vessel size, and a design of the vessel. Examples of the operating conditions may include, but are not limited to, a minimum wind speed required for a propulsion of the vessel, a maximum wind speed tolerable for a safe operation of the vessel, an optimal range of wind speeds for efficient propulsion, a maximum wave height for safe operation, an impact of wind propulsion on vessel speed and course, a manoeuvrability in different wind conditions, effects on navigation and control systems in the vessel, a temperature affecting equipment and materials in the vessel, a corrosion resistance in marine environments, impact of salt spray and moisture on components, a compatibility of wind propulsion system with a vessel structure, an aerodynamic or hydrodynamic performance under different conditions, a structural integrity under load from wind forces, emergency procedures for loss of wind or propulsion failure, and environmental regulations and emissions standards. The data of the at least one operational parameter may provide information pertaining to a current state and an operational behavior of the at least one of: the at least one wind propulsion device, the vessel. In this regard, the at least one of: the at least one wind propulsion device, the vessel, is equipped with operational sensors that are designed to measure the data of the at least one operational parameter. Furthermore, such operational sensors may be used to generate the data based on measurements captured during different operating conditions. Optionally, the data of the at least one operational parameter is determined using one or more second sensors, and wherein the data of the at least one operational parameter comprises at least one of: wind speed and direction data, wind-capturing component data, weather forecast data, vessel movement data, vessel position data, vessel manoeuvring data, vessel-engine data.

[0034] The term "second sensor" refers to a device that is capable of detecting and / or measuring at least one operational parameter under a given operating condition associated with the at least one of: the at least one wind propulsion device, the vessel. Optionally, at least some of the one or more second sensors are arranged on the at least one wind propulsion device and / or the vessel. It will be appreciated that some second sensors may be arranged on the at least one wind propulsion device, and remaining second sensors may be arranged on other different parts of the vessel. Different types of the one or more second sensors may be employed for measuring different types of operational parameters (as discussed hereinabove).

[0035] Optionally, the one or more second sensors are operatively mounted on the vessel or remotely communicably coupled to the vessel. In this regard, the one or more second sensors may be physically installed on the at least one wind propulsion device and / or various parts of the vessel. Moreover, when the one or more second sensors are arranged on the at least one of: the at least one wind propulsion device, the vessel itself, it is ensured that a direct and immediate measurement of the data of the at least one operational parameter is obtained as real-time data, that could be beneficial for decision-making and control strategies. Alternatively, the one or more second sensors are remotely communicably coupled to the at least one of: the at least one wind propulsion device, the vessel, via, for example, well-known wireless technologies and communication protocols. Moreover, the one or more second sensors may use a telemetry arrangement (such as a satellite communication network, a cellular network, or similar) to transmit data to the processor. A technical benefit of remotely placing the one or more second sensors is to reduce a need for physical connections, minimizing a potential interference with the vessel's operation.

[0036] The term "wind speed and direction data" refers to information pertaining to a wind speed and a wind direction. In other words, the wind speed and direction data comprises information related to a flow speed at which the wind is blowing and in which direction it is blowing, when the vessel is in operation. The wind speed can be measured, for example, in terms of meters per second (m / s), kilometres per hour (km / h), miles per hour (mph), knots (kt or kn), feet per second (ft / s), and the like. Additionally, the wind direction may be measured clockwise from a north direction, where 0 degree or 360 degrees correspond to the north direction, 90 degrees correspond to an east direction, 180 degrees to correspond to a south direction, and 270 degrees correspond to a west direction. For example, 10 m / s and 270 degrees could mean that a wind having a speed of 10 meters per second is coming from the west direction. Examples of the one or more second sensors when employed to measure the wind speed and direction data may include, but are not limited to, an anemometer, a wind vane, and an ultrasonic wind measurement device.

[0037] The term "weather forecast data" refers to predictive information pertaining to meteorological conditions such as wind speed, wind direction, sea state, temperature, and the like. This data helps vessel operators and wind propulsion system designers anticipate and plan for optimal sailing conditions, enabling efficient utilization of wind energy for propulsion while ensuring safety and operational effectiveness. Furthermore, the weather forecast data could also comprises data pertaining to at least one of: temperature forecasts, precipitation forecasts, wind forecasts, humidity forecasts, and atmospheric pressure forecasts. The weather forecast data allows the at least one of: the at least one wind propulsion device, the vessel to proactively adjust its configuration in preparation for higher wind speeds. The weather forecast data may provide information about expected weather conditions, helping the at least one wind propulsion device to anticipate and adapt to changes according to environmental factors. It should be noted that, the weather forecast data may not be directly measured by onboard sensors (namely, those second sensors that are arranged in the vessel itself), but may be received from external sources, for example, weather stations, satellites, and other specialized weather monitoring systems.

[0038] The term "vessel movement data" refers to information pertaining to a motion and a behavior of the vessel in response to wind and other environmental factors. The vessel movement data may comprise at least one of: a vessel speed, a vessel heading, a vessel roll angle, a vessel pitch angle, a vessel course, a vessel drift, a vessel stability. The aforesaid terms are well-known in the art. Examples of the one or more second sensors when employed to measure the vessel movement data may include, but are not limited to, an inertial measurement unit (IMU), an accelerometer, an inclinometer, a global positioning system (GPS), a global navigation satellite system (GNSS) a gyroscope, a magnetic compass sensor.

[0039] The term "vessel position data" refers to information pertaining to a geographical location of the vessel at a given time. The vessel position data may provide geographical coordinates of the vessel at a specific point in time, which may comprise the vessel's latitude, longitude, and in some cases altitude. The vessel position data is typically determined using the GPS or other satellite navigation systems.

[0040] The term "vessel manoeuvring data" refers to information pertaining to a movement of the vessel during its navigation. The vessel manoeuvring data may provide information related to a dynamic movement of the vessel, especially in terms of changes in its course, speed, and other manoeuvres. The vessel manoeuvring data may also provide information about how the vessel is responding to control inputs, environmental conditions, and navigational requirements. Specifically, the vessel manoeuvring data may provide an idea about vessel's steering mechanism. Beneficially, the vessel manoeuvring data may facilitates in assessing effectiveness of the at least one wind propulsion device, optimising sail configurations, and ensuring safe and efficient manoeuvring in various environmental conditions. Optionally, the vessel manoeuvring data comprises at least one of: a rudder angle, a vessel engine load, a vessel speed, a vessel heading, a vessel stability-related data. It will be appreciated that when the data of the at least one operational parameter comprising at least the vessel manoeuvring data, is served as an input for training the Al model, it can accurately predict a relationship between the performance of the at least one wind propulsion device and the different control inputs, and can therefore help in optimizing an operation of the vessel in real-world maritime scenarios.

[0041] The term "vessel-engine data" refers to information pertaining to an operation and a performance of at least one engine arranged on the vessel. The vessel-engine data may comprise values of at least one of: a propulsion power of the vessel-engine, a shaft torque of the vesselengine, a speed of the vessel-engine, a fuel consumption by the vesselengine, engine health diagnostics. In this regard, examples of the one or more second sensors employed for measuring the vessel-engine data may include, but are not limited to, power transducers, torque sensors, load sensors, tachometers, fuel flow meters, temperature sensors, pressure sensors, exhaust gas temperature sensors, coolant level sensors, oil level sensors, voltage sensors, hour meters, emissions sensors, vibration sensors, speed sensors, engine load sensors, alternator output sensors, and oil quality sensors. It will be appreciated that the vessel-engine data provides benefits such as optimizing performance, planning maintenance, improving fuel efficiency, enhancing safety, ensuring compliance, aiding decision-making, tracking performance, mitigating risks, managing assets, and the like.

[0042] The term "wind-capturing component" refers to a device that is capable of capturing wind energy. Examples of wind-capturing components may include, but are not limited to, sails, rotors, or any other structures intended to interact with the wind energy for the propulsion of the vessel. Herein, the term "wind-capturing component data" refers to information pertaining to a performance and a behaviour of a wind-capturing component that is responsible for capturing the wind energy. The windcapturing component data may include parameters such as a wind speed, a wind direction, a wind pressure, aerodynamic forces acting on the windcapturing component, an angle of attack, and an efficiency of energy conversion. Therefore, optionally, the wind capturing-component data is associated with one or more wind-capturing components. It will be appreciated that by analysing data from each wind-capturing component, adjustments could be made to optimize configurations and operations of the one or more wind-capturing components. This may ensure that the vessel achieves maximum propulsion efficiency under varying wind conditions. This may enable operators to adjust angles of attack, trim settings, or other parameters to fine-tune performance and maintain stability of the vessel. Moreover, analysing the wind-capturing component data may enable in improved energy conversion and utilization, ultimately maximizing the propulsion output of the wind propulsion system. Optionally, the data of the at least one operational parameter comprises relative onboard wind data. Herein, the term "relative onboard wind data" refers to information pertaining to wind conditions as experienced onboard the vessel and the at least one wind propulsion device (namely, when the wind is blowing and the vessel is also in operation i.e., sailing on the water body). It may be appreciated that relative onboard wind data refers to information pertaining to the wind conditions experienced by the at least one wind propulsion device, in real-time, relative to a movement of the vessel which is sailing on the water body. Notably, the movement of the vessel may be in same direction in which the wind flows, may be in a direction opposite to the direction in which the wind flows, or in an angular manner to the direction in which the wind flows. It may be appreciated that relative onboard wind data provides information on wind conditions experienced by the at least one wind propulsion device relative to the wind speed, the wind direction, and the movement direction of the vessel. It may be appreciated that the relative onboard wind data is used for assessment of performance of the wind propulsion device during such wind conditions. In this regard, the relative onboard wind data is measured with respect to a position and a movement of the vessel and the at least one wind propulsion device, providing information on how the wind affects a movement of the vessel and the at least one wind propulsion device. It is to be noted that the relative onboard wind data is not an absolute measure of wind conditions when the vessel may be stationary in the water body, but rather reflects wind conditions when the vessel is in motion. The relative onboard wind data is obtained by means of using sensors arranged on the vessel and / or the at least one wind propulsion device. It will be appreciated that the relative onboard wind data provides a real-time check on the wind speed and the wind direction in various wind conditions.

[0043] Optionally, the data of the at least one operational parameter comprises vessel electrical load balance and generator capacity data. In this regard, a distribution and a balance of electrical loads on an electrical arrangement of the vessel may include all devices that consume an electrical power available on the vessel, such as lighting, navigation equipment, communication systems, and other electronic components. Typically, monitoring the electrical load balance may provide information pertaining to how much electrical power is being consumed by the vessel. Moreover, the generator capacity data provides information pertaining to a capacity of an electrical generator of the vessel. The electrical generator is responsible for producing the electrical power on the vessel, and its capacity denotes a maximum amount of electrical power it can supply. Beneficially, the generator capacity data may ensure that the vessel movement is stable. It will be appreciated that when the vessel electrical load balance and generator capacity data (along with other types of the data of the at least one operational parameter, as discussed earlier) is used to train the Al model, the trained Al model gains an ability to predict how changes in electrical load, generator capacity, may impact the performance of the at least one wind propulsion device. Beneficially, due to this, operation(s) of the at least one wind propulsion device could be adjusted accordingly, to maximize energy efficiency of the at least one wind propulsion device, and ensure that an available electrical power is used judiciously.

[0044] Optionally, the data of the at least one operational parameter comprises route data and hindcast weather data. The term "route data" refers to information pertaining to a path of the vessel during its navigation. Optionally, the route data pertains to at least one of: a historical route, a planned route, a current route, of the vessel. It will be appreciated that the route data comprises information related to waypoints, course headings, distances between waypoints, expected arrival times, navigational hazards along a route, and the like. It will also be appreciated that when the route data is fed into the Al model for predicting the performance of the at least one wind propulsion device, the route data facilitates in providing suggestions to minimize fuel consumption, reduce voyage duration, and enhance overall operational efficiency of the at least one wind propulsion device. By following an optimized route, vessels can avoid adverse weather conditions, rough seas, and congested areas, leading to cost savings and improved safety. Furthermore, the term "hindcast weather data" refers to historical weather data reconstructed for past time periods. The hindcast weather data comprises parameters, for example, such as a wind speed, a wind direction, a sea state, an atmospheric condition, for said past time periods. It will be appreciated that when the hindcast weather data (combined with other types of data as discussed earlier) is fed into the Al model for predicting the performance of the at least one wind propulsion device, the hindcast weather data enhances its predictive accuracy. This may be because by training the Al model with comprehensive historical weather data, the Al model could learn patterns and correlations that contribute to more accurate performance predictions. This may allow operators to anticipate and optimise a performance of the at least one wind propulsion device based on predictions from the Al model, leading to more efficient vessel operation and improved decision-making.

[0045] Notably, upon receiving the force measurement data and the data of the at least one operational parameter, the processor is configured to feed the aforesaid data into the Al model for training it, in order to establish the predictive relationship between the performance of the at least one wind propulsion device and the different control inputs, upon the Al model has been trained. Optionally, the different control inputs comprise at least the force measurement data and the data of the at least one operational parameter.

[0046] It may be appreciated that the term "predictive relationship" refers to information how the performance of the at least one wind propulsion device is affected by different control inputs, based on the force measurement data and the data of the at least one operational parameter. For example, a first performance of the at least one wind propulsion device is noted when the control inputs are set at a first value, and a second performance of the at least one wind propulsion device is noted when the control inputs are set at a second value. The first and the second performances are analysed and a predictive relationship between the performance and the different control inputs is established.

[0047] It will be appreciated that training the Al model involves using the force measurement data and the data of the at least one operational parameter to teach the Al model to recognize patterns and establish relationships between the different control inputs and a target output (namely, the performance of the at least one wind propulsion device). In other words, the force measurement data and the data of the at least one operational parameter are provided by the processor to Al model as a training dataset. This training dataset would cover a diverse range of scenarios, representing various combinations of wind conditions and operating conditions of the vessel. Furthermore, said training data may be pre- processed by handling missing values, outliers, and any inconsistencies, and may be normalized into numerical features to bring them into a consistent scale, which helps the Al model to converge faster during training. It is to be understood that during training, relevant features (in the force measurement data and the data of the at least one operational parameter) that contribute significantly to predicting the performance of the at least one wind propulsion device may be identified, and any irrelevant or redundant features that may not contribute meaningfully to the prediction are excluded from the training dataset. Furthermore, the training dataset may be split into a training set and a testing set, wherein the training set is used to teach the Al model, while the testing set is reserved to evaluate the Al model's performance on new, unseen data. The Al model's architecture may be chosen for the prediction, such as regression models, decision trees, ensemble methods, or more advanced models like neural networks, and the training dataset is fed into the chosen the Al model. Throughout a training process, the Al model learns to establish the predictive relationship. Such a predictive relationship may allow the Al model to predict the performance of the at least one wind propulsion device under different conditions and with different control inputs. Once the Al model is trained, it is evaluated and validated to ensure its accuracy and generalization ability. The Al model could also be referred to as a machine learning model or a deep learning model.

[0048] It may be appreciated that use of the force measurement data and the data of the at least one operational parameter for training the Al model, results in the technical effect of improved adaptability, data richness, and accuracy of the Al model, in establishing the predictive relationship between the performance of the at least one wind propulsion device with different control inputs.

[0049] In an example, given a wind speed of X m / s measured from ship's bridge, the Al model may predict the sail force generated by various control inputs, considering factors such as a rotational speed in a case of a rotor sail. This prediction is based on the Al model's training with sail force data across various wind conditions, leveraging principles of the Magnus effect. Herein, the advantage lies in the Al model's ability to learn and predict sail force accurately under real-world airflow disturbances, thereby enhancing realism in prediction. In another example, when ship's course is X, the Al model may forecast the sail force resulting from various control inputs, accounting for a change in course by Y degrees. This change impacts relative wind's direction, speed, and disturbances affecting the sails. By incorporating knowledge of actual wind conditions, the Al model better comprehends how course adjustments influence sail force production. In yet another example, for a ship speed of X knots, the Al model may predict a sail force produced with control inputs when the ship's speed changes from X knots to Y knots. A correlation between the ship speed and sail power predominantly stems from alterations in relative wind's characteristics (such as, a direction and a speed) on the sails. With insights into real wind dynamics, the Al model can accurately gauge effects of changes in ship speed on sail force.

[0050] Optionally the method further comprises training the artificial intelligence model with computational fluid dynamics data associated with the at least one of: the at least one wind propulsion device, the vessel. Herein the term "computational fluid dynamics data" refers to information obtained through simulations of fluid flow around the at least one wind propulsion device and the vessel. In this regard, the computational fluid dynamics data may be obtained using a simulation model that represents the at least one wind propulsion device, the vessel, the operating condition associated with at least one of: the at least one wind propulsion device, the vessel as well as the wind conditions in real-time. For example, to obtain the computational fluid dynamics data on the wind propulsion device, the simulation with a suitable platform (such as, Matlab®) may be performed by the following steps: (i) the wind propulsion device geometry (such as height, width, length, area of wind contact) is modelled (preferably, in a 3 dimension modelling); (ii) boundary conditions such as the wind speed, wind direction, wind pressure and so on are set (based on real-time data received from at least one of: the first sensor, the second sensor); (iii) air around the wind propulsion device, is divided into a fine mesh of small virtual cells to capture the detailed flow characteristics (such as turbulence of the wind flow) by using relevant mathematical formulae, for example but not limited to, Navier-Stokes equations; and (iv) iteratively solving the relevant mathematical formulae until a reliable solution is obtained (i.e., until changes between iterations are minimal). Afterwards, from the solution the computational fluid dynamics data is obtained. It may be appreciated that the computational fluid dynamics data may also provide information on the wind speed, the wind direction, areas of high and low wind pressure and so on. The simulation of fluid flow around the at least one wind propulsion device and the vessel is obtained under different control inputs. In this regard, the computational fluid dynamics data may provide a detailed performance of the at least one wind propulsion device by modelling / simulating the airflow around its components (such as Magnus rotors or aerofoil sails). The computational fluid dynamics data may include information on lift and drag forces, airflow velocities, and pressure distributions. Furthermore, the Al model is trained using the computational fluid dynamics data in addition to the force measurement data and the data of the at least one operational parameter. During training, the Al model may learn relationships between a combined set of features and a target output (namely, the performance of the at least one wind propulsion device) under various conditions. Beneficially, such an approach may allow the Al model to establish a relationship between design parameters, operating conditions, and resulting performance of the at least one wind propulsion device, contributing to highly accurate and informed predictions for real-world scenarios.

[0051] Optionally, the method further comprises utilising the predictive relationship to determine at least one wind propulsion device control parameter. In this regard, in addition to predicting the at least one wind propulsion performance, the Al model's forecast can be leveraged to optimize the at least one wind propulsion device control parameters. The term "control parameters" refers to different control inputs pertaining to the wind-capturing components (such as a sail). For example, the Al model may predict the force produced by a sail under varying wind speeds, ship courses, and ship speeds. Based on these predictions, adjustments to sail control inputs such as sail angle, rotation speed (for rotor sails), or deployment tactics can be dynamically optimised to maximize an efficiency and effectiveness of the at least one wind propulsion system. Additionally, the artificial model can predict an impact of the sail force on a ship tilt (for example with the force produced by the sail X, the ship tilts by an average of Y, and the Al model predicts what the average ship tilt will be like with all possible operating inputs in a current situation) and electrical power consumption (for example, when operating the sail in the current situation consumes X kW of electricity and produces Y kN of power, the Al model predicts what power generated by the sail will be if the available electrical power is limited or if more electricity is available). This enables in providing informed decisions regarding power management and sail operation in the vessel. This beneficially enables vessel operators to make informed decisions in realtime, ensuring optimal performance and fuel savings while navigating varying environmental conditions.

[0052] Optionally, forecast provided by the Al model can be leveraged to optimize the at least one wind propulsion device control parameters to improve performance of the at least one wind propulsion device. As a result, added information pertaining to the force measurement data and the data of at least one operational parameter may be generated and such generated information may facilitate data richness for training the Al model further, in order to improve adaptability, and forecast accuracy. Moreover, the predictive relationship between the performance of the at least one wind propulsion device and different control inputs allows the Al model to accurately predict the performance of the at least one wind propulsion device with different control inputs.

[0053] The present disclosure also relates to the system as described above. Various embodiments and variants disclosed above, with respect to the aforementioned method, apply mutatis mutandis to the system.

[0054] Optionally, the at least one wind propulsion device comprises: one or more wind-capturing components, operatively coupled to the processor, for generating thrust using the force generated by wind condition on the vessel; and power transmission mechanism operatively coupled to the one or more wind-capturing components and the processor.

[0055] In this regard, the wind-capturing components, such as magnus-rotors or aerofoil sails, are designed to capture the energy from the wind. When the wind flows over these components, it imparts a force on them, creating lift or thrust. The wind-capturing components are strategically positioned on the vessel to harness the available wind energy. The processor receives data or information from the wind-capturing components. The data may include measurements related to the forces generated by the wind-capturing components, as well as other relevant information about their status and performance. Based on the data received, the processor can optimize the configuration or orientation of the wind-capturing components to maximize thrust or efficiency. This optimisation may involve adjusting the angle of the wind-capturing components or other parameters to align with the prevailing wind conditions. Further, the power transmission mechanism transfers a mechanical energy generated by the wind-capturing components to other components on the vessel. The power transmission mechanism then translates the captured wind energy into a form that is suitable for the propulsion of the vessel. This may involve converting the rotational motion of the wind-capturing components into mechanical energy that can drive the vessel forward.

[0056] Optionally, in the system, the data of the at one operational parameter comprises relative onboard wind data. Optionally, in the system, the data of the at least one operational parameter comprises vessel electrical load balance and generator capacity data.

[0057] Optionally, in the system, the data of the at least operational parameter comprises route data and hindcast weather data.

[0058] Optionally, in the system, the force measurement data comprises at least one of: pressure-based force measurement data, strain-based force measurement data, displacement-based force measurement data.

[0059] Optionally, in the system, the data of at least one operational parameter comprises at least one of: wind speed and direction data, wind-capturing component data, weather forecast data, vessel movement data, vessel position data, vessel manoeuvring data, vessel-engine data.

[0060] Optionally, in the system, the wind-capturing component data is associated with one or more wind-capturing components.

[0061] Optionally, in the system, the one or more second sensors are operatively mounted on the vessel or remotely communicably coupled to the vessel.

[0062] Optionally, in the system, the processor is further configured to train the artificial intelligence model with computational fluid dynamics data associated with the at least one of: the at least one wind propulsion device, the vessel. Optionally, in the system, the processor is further configured to utilise the predictive relationship to determine at least one wind propulsion device control parameter.

[0063] DETAILED DESCRIPTION OF THE DRAWINGS

[0064] Referring to FIG. 1, illustrated are steps of a method for training an Al model for predicting a performance of the at least one wind propulsion device on a vessel, in accordance with an embodiment of the present disclosure. At step 102, force measurement data is received, wherein the force measurement data is indicative of a force generated by the at least one wind propulsion device. At step 104, data of at least one operational parameter is received, wherein the data of the at least one operational parameter is indicative of an operating condition associated with the at least one of: the at least one wind propulsion device, the vessel. At step 106, the Al model is trained with the force measurement data and the data of the at least one operational parameter to establish a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

[0065] The aforementioned steps are only illustrative and other alternatives can also be provided where one or more steps are added, one or more steps are removed, or one or more steps are provided in a different sequence without departing from the scope of the claims.

[0066] Referring to FIG. 2, illustrated is an exemplary scenario in which a system 200 for training an Al model for predicting a performance of at least one wind propulsion device (depicted as a wind propulsion device 202) on a vessel 204, is employed, in accordance with an embodiment of the present disclosure. With reference to FIG. 2, the system 200 is being used in the vessel 204 comprising the wind propulsion device 202. The system 200 comprises one or more first sensors (for example, depicted as first sensors 206a and 206b), one or more second sensors (for example, depicted as second sensors 208a and 208b), and a processor 210. The processor 210 is operatively coupled to the first sensors 206a- b and the second sensors 208a-b. The wind propulsion device 202 comprises one or more wind-capturing components (depicted as a wind- capturing component 212), and a power transmission mechanism 214.

[0067] The power transmission mechanism 214 is operatively coupled to the wind-capturing component 212 and the processor 210. The processor 210 is configured to perform various operations, as described earlier with respect to the aforementioned first aspect. FIG. 2 is merely an example, which should not unduly limit the scope of the claims herein. A person skilled in the art will recognize many variations, alternatives, and modifications of embodiments of the present disclosure.

Claims

CLAIMS1. A method for training an artificial intelligence model for predicting a performance of at least one wind propulsion device (202) on a vessel (204), wherein the method comprises: receiving, by a processor (210), force measurement data indicative of a force generated by the at least one wind propulsion device; receiving, by a processor, data of at least one operational parameter indicative of an operating condition associated with at least one of: the at least one wind propulsion device, the vessel; and training, by a processor, the artificial intelligence model with the force measurement data and the data of the at least one operational parameter for establishing a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

2. A method according to claim 1, wherein the data of the at least one operational parameter comprises relative onboard wind data.

3. A method according to claim 1, wherein the data of the at least one operational parameter comprises vessel electrical load balance and generator capacity data.

4. A method according to claim 1, wherein the data of the at least operational parameter indicative of operating condition associated with the vessel (204) is the vessel location data, optimized route data and hindcast weather data.

5. A method according to any of the preceding claims, wherein the force measurement data is determined using one or more first sensors (206a, 206b), and wherein the force measurement data comprises at least one of:pressure-based force measurement data, strain-based force measurement data, displacement-based force measurement data.

6. A method according to any of the preceding claims, wherein the data of the at least one operational parameter is determined using one or more second sensors (208a, 208b), and wherein the data of the at least one operational parameter comprises at least one of: wind speed and direction data, wind-capturing component data, weather forecast data, vessel movement data, vessel position data, vessel manoeuvring data, vessel-engine data.

7. A method according to claim 6, wherein the wind-capturing component data is associated with one or more wind-capturing components (212).

8. A method according to claim 6 or 7, wherein the one or more second sensors (208a, 208b) are operatively mounted on the vessel (204) or remotely communicably coupled to the vessel.

9. A method according to any of the preceding claims, further comprising training the artificial intelligence model with computational fluid dynamics data associated with the at least one of: the at least one wind propulsion device (202), the vessel (204).

10. A method according to any of the preceding claims, further comprising utilising the predictive relationship to determine at least one wind propulsion device control parameter.

11. A system (200) for training an artificial intelligence model for predicting a performance of at least one wind propulsion device (202) on a vessel (204), the system comprising: one or more first sensors (206a, 206b) for determining force measurement data indicative of a force generated by the at least one wind propulsion device; one or more second sensors (208a, 208b) for determining data of at least one operational parameter indicative of an operating condition associated with at least one of: the at least one wind propulsion device, the vessel; and a processor (210) operatively coupled to the one or more first sensors and the one or more second sensors, wherein the processor is configured to: receive the force measurement data; receive the data of the at least one operational parameter; and train the artificial intelligence model with the force measurement data and the data of the at least one operational parameter for establishing a predictive relationship between the performance of the at least one wind propulsion device and different control inputs.

12. A system (200) according to claim 11, wherein the at least one wind propulsion device (202) comprises: one or more wind-capturing components (212), operatively coupled to the processor (210), for generating thrust using the force generated by the at least one wind propulsion device; and power transmission mechanism (214) operatively coupled to the one or more wind-capturing components and the processor.

13. A system (200) according to claim 11 or 12, wherein the force measurement data comprises at least one of: pressure-based force measurement data, strain-based force measurement data, displacement-based force measurement data.

14. A system (200) according to any of claims 11-13, wherein the data of at least one operational parameter comprises at least one of: wind speed and direction data, wind-capturing component data, weather forecast data, vessel movement data, vessel position data, vessel manoeuvring data, vessel-engine data.

15. A system (200) according to any of claims 11-14, wherein the processor (210) is further configured to utilise the predictive relationship to determine at least one wind propulsion device control parameter.

Images