Estimation of the sloshing response of a tank using a statistical model trained by machine learning

Abstract

The present invention relates to the estimation of the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas. A statistical model is trained using a supervised machine learning method based on a test data set, which may include sea test data, the statistical model being capable of estimating the sloshing response of the tank based on the filling level of the tank and the current sea state and optionally at least one of the following: the draft of the ship, the speed of the ship or the heading of the ship. The statistical model trained in this way is used to estimate the sloshing response of a sealed and thermally insulated tank (2) for transporting liquefied gas. In an alternative embodiment, the statistical model estimates the sloshing response based on the filling level of the tank and the current sea state and optionally based on at least one of the following: the draft of the ship, the speed of the ship or the heading of the ship.

Description

Technical Field

[0001] The present invention relates to the estimation of the sloshing response of a sealed and thermally insulated tank for transporting liquefied gases. More specifically, the present invention relates to a method for obtaining a statistical model capable of estimating the sloshing response of such a tank, a method for obtaining a database that can be used to estimate the sloshing response of such a tank, and a management system for a ship comprising at least one such tank. In this document, the term "ship" refers to a vehicle that transports and / or uses its own propulsion to transport liquefied gases between two points on the earth, or a floating processing and / or storage unit for one or more liquefied gases. As a standard practice for ships using liquefied gases as fuel, LNG propulsion ships, container ships, cruise ships, and bulk carriers may be mentioned. Background Art

[0002] Sealed and thermally insulated tanks are generally used for storing and / or transporting liquefied gases at low temperatures, such as tanks for transporting liquefied petroleum gas (also known as LPG) at temperatures between, for example, -50°C and 0°C (including the endpoints), or for transporting liquefied natural gas (LNG) at approximately -162°C at atmospheric pressure. These tanks can be designed for transporting liquefied gases and / or receiving liquefied gases, which are used as fuel to propel floating structures. Multiple liquefied gases can also be envisaged, especially methane, ethane, propane, butane, ammonia, dihydrogen, or ethylene.

[0003] Ship tanks can be single-sealed membrane tanks or double-sealed membrane tanks that allow transportation at atmospheric pressure. The sealing membrane is generally made of thin stainless steel or invar sheets. The membrane usually comes into direct contact with the liquefied gas.

[0004] During its transportation, the liquid contained in the tank is subject to various movements. In particular, the movement of the ship at sea, for example due to climatic conditions such as sea state or wind, which causes the agitation of the liquid in the tank. The agitation of the liquid (commonly referred to as sloshing) generates stresses on the tank walls, which may damage the integrity of the tank.

[0005] These sloshing phenomena occur on natural gas (hereinafter referred to as LNG) transport and / or user vessels (commonly referred to as "LNG as fuel" vessels) or methane tankers, as well as on moored storage vessels called FPSO (Floating Production Storage and Offloading) vessels, which moored storage vessels serve, for example, extraction platforms and natural gas liquefaction units, commonly referred to as FLNG (Floating Liquefied Natural Gas) units or FSRU (Floating Storage and Regasification Units), that is, more generally floating production, storage and export support. The sloshing phenomena occur equally in agitated sea conditions and in almost calm sea conditions if the liquefied gas cargo resonates with the excitation generated by even low surges to which the vessel is subjected. In these resonance cases, the sloshing can become extremely violent, especially in the case of wave breaking on vertical walls or in corners, and there is thus a risk of deterioration of the liquefied gas containment system or of the insulation system located just behind said containment system.

[0006] Now, in the context of liquefied gas tanks such as LNG tanks, the integrity of the tank is of particular importance due to the flammable or explosive nature of the liquid being transported and the risk of cold spots on the steel hull of the floating unit in the event of a leak.

[0007] U.S. Patent 8,643,509 discloses a method for reducing the risk associated with the sloshing of a liquefied gas cargo. In this document, the resonance frequency of the liquid in the tank is estimated based on the tank and its filling level. During transportation, the movement frequency of the vessel is evaluated according to the climatic and ocean conditions as well as the speed of the vessel. The predicted movement frequency is also evaluated according to the route that the vessel is to follow. If any movement frequency becomes too close to the resonance frequency of the liquid in the tank, an alarm is issued to change the route and / or change the speed of the vessel to avoid a dangerous situation.

[0008] Despite the measures provided to reduce sloshing, the sloshing of the liquid in the tank, especially due to resonance phenomena, can lead to the following local risks: deformation of the primary seal primary membrane, damage to the underlying structure in the primary and / or secondary space where the primary seal membrane is located, falling objects - especially objects falling from static equipment - that can damage the primary seal membrane in the short or medium term, or more generally deform the primary seal membrane beyond its structural tolerances.

[0009] Therefore, there is still a need for a method for estimating the sloshing response of a tank during vessel navigation and, if necessary, applying the necessary measures to prevent excessive sloshing that poses a risk of damaging the primary seal membrane of the tank. Summary of the Invention

[0010] One idea behind the present invention is to use supervised machine learning methods to train a statistical model that can estimate the sloshing response of a tank based on the filling level of the tank, the current sea state, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship. The statistical model is trained based on a test data set obtained from the results of a plurality of tests, each test including: subjecting a test tank having a given filling level to movement; and measuring the pressure at at least one point on the wall of the test tank and / or the number of impacts on at least one wall of the test tank. Thereafter, the statistical model can be used to estimate the sloshing response of the tank in the framework of a management system for a ship, for example, by constructing a database that can be used for real-time consultation.

[0011] According to an embodiment following a first variant, the present invention provides a method for obtaining a statistical model that can estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas, the method comprising the following steps, the steps comprising:

[0012] - Training a statistical model by a supervised machine learning method based on a test data set, the statistical model being able to estimate the sloshing response of a tank based on the filling level of the tank, the current sea state, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship, and the test data set being obtained from the results of a plurality of tests, each test including: subjecting a test tank having a given filling level to movement; and measuring the pressure at at least one point on the wall of the test tank and / or the number of impacts on at least one wall of the test tank.

[0013] According to an embodiment following a second variant, the present invention further provides a method for obtaining a statistical model that can estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas, the method comprising the following steps, the steps comprising:

[0014] - Training a statistical model by a supervised machine learning method based on a test data set, the statistical model being able to estimate the sloshing response of the tank based on the filling level of the tank, the current movement state of the ship, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship, and the test data set being obtained from the results of a plurality of tests, each test including: subjecting a test tank having a given filling level to movement; and measuring the pressure at at least one point on the wall of the test tank and / or the number of impacts on at least one wall of the test tank.

[0015] The so-called "supervised machine learning method" means a machine learning (also called artificial learning or statistical learning in France) method that learns a prediction function based on annotated examples. In other words, the supervised machine learning method enables the construction of a model that can make predictions based on multiple examples where the response to be predicted is known. The supervised machine learning method is typically executed by a computer; thus, the step of training a statistical model is typically executed by a computer.

[0016] In the context of the present invention, the statistical model to be trained is capable of estimating the sloshing response including one or more quantitative variables based at least on the filling level of the tank and the current sea state or the current movement state of the ship. Therefore, the statistical model to be trained is capable of solving a regression problem.

[0017] Since it includes the step of training a statistical model based on a test data set by a supervised machine learning method, the statistical model is capable of estimating the sloshing response of the tank by calculation based at least on the filling level of the tank and the current sea state or the current movement state of the ship (including the values of the filling levels of the tanks not tested and the current sea state and the movement state of the ship). Therefore, the statistical model can be used to estimate the sloshing response of the tank under the actual use conditions on the ship.

[0018] The test tank can be smaller than the tank whose sloshing response is to be estimated by the statistical model. It can have a geometry representative of the tank whose sloshing response is to be estimated by the statistical model. In addition, it is obvious that in the phrase or feature "training a statistical model based on a test data set by a supervised method", the "test data set" can also include or contain data from so-called "real" activities, i.e., data obtained or measured on a ship serving as a liquefied gas carrier ship and / or a user ship.

[0019] According to an embodiment, the method described above can include one or more of the following features.

[0020] The so-called sloshing response means any parameter and set of qualitative and / or quantitative parameters that can represent the mechanical load applied to the tank during the cargo sloshing phenomenon.

[0021] According to one embodiment, the sloshing response includes at least one of the following: the number of impacts of the fluid on the tank wall, the maximum pressure on the wall of the tank, and the likelihood of damaging the tank.

[0022] Therefore, the statistical model is capable of estimating the likelihood of damage to the tank and / or the parameters that enable the estimation of such likelihood of damage.

[0023] According to one embodiment, the supervised machine learning method is a Gaussian process regression method.

[0024] The Gaussian process regression method is well-suited for training statistical models because it can train based on a relatively limited amount of data to produce a statistical model that can solve regression problems for any input dataset. However, other supervised machine learning methods can be employed without departing from the scope of the present invention.

[0025] According to one embodiment, at least one constraint is imposed on the statistical model during the training of the statistical model by a supervised machine learning method.

[0026] Thus, the training of the statistical model can be defined based on basic physical considerations such as no sloshing when the filling level of the tank is zero and / or based on considerations obtained through practical experience such as the fact that the greater the movement of the tank or the greater the size of the tank may potentially result in a higher sloshing response. As a result, the accuracy of the statistical model's estimation of the sloshing response is improved.

[0027] According to one embodiment, the method further includes the step of excluding from the test dataset test results characterized by a sloshing response below a threshold before the step of training the statistical model.

[0028] Thus, the statistical model is trained only based on test data that reveals severe sloshing, especially in terms of the number of impacts. In fact, according to one aspect of the present invention, the number of events encountered is a more important factor for statistical convergence than the intensity of the impact. As a result, the accuracy of the statistical model's estimation of the sloshing response is further improved.

[0029] According to one embodiment, the statistical model considers multiple tanks, and the statistical model is capable of estimating the sloshing response of each tank based on the position of each tank in the ship.

[0030] According to one embodiment following the first variant, the present invention also provides a system for obtaining a statistical model that can estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas. The system includes processing means configured to train the statistical model based on a test dataset by a supervised machine learning method. The statistical model is capable of estimating the sloshing response of the tank based on the filling level of the tank and the current sea state and optionally at least one of the following: the draft of the ship, the speed of the ship, and the course of the ship. And the test dataset is obtained from the results of multiple tests, each test including: subjecting a test tank with a given filling level to movement; and measuring the pressure at at least one point on the wall of the test tank and / or the number of impacts on at least one wall of the test tank.

[0031] According to an embodiment following the second variant, the present invention further provides a system for obtaining a statistical model capable of estimating the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas. The system includes a processing device configured to train the statistical model based on a test data set by a supervised machine learning method. The statistical model is capable of estimating the sloshing response of the tank according to the filling level of the tank, the current movement state of the ship, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the course of the ship. And the test data set is obtained from the results of a plurality of tests, each test including: subjecting a test tank with a given filling level to movement; and measuring the pressure at at least one point on the wall of the test tank and / or the number of impacts on at least one wall of the test tank.

[0032] Such a system achieves the same advantages as the method described above.

[0033] According to an embodiment following the first variant, the present invention further provides a method for obtaining a database that can be used to estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas. The method includes the following steps, which include:

[0034] - Generating a plurality of input data vectors, each input data vector including the filling level of the tank and the current sea state; and

[0035] - For each input data vector generated in this way: obtaining the estimated sloshing response of the tank by means of the statistical model obtained by the above method following the first variant, and storing the estimated sloshing response of the tank associated with the input data vector in the database.

[0036] According to an embodiment following the second variant, the present invention further provides a method for obtaining a database that can be used to estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas. The method includes the following steps, which include:

[0037] - Generating a plurality of input data vectors, each input data vector including the filling level of the tank and the current movement state of the ship; and

[0038] - For each input data vector generated in this way: obtaining the estimated sloshing response of the tank by means of the statistical model obtained by the above method following the second variant, and storing the estimated sloshing response of the tank associated with the input data vector in the database.

[0039] Although the statistical model can estimate the sloshing response of the tank by calculation for the filling level values of untested tanks and the current sea state or the moving state of the ship, the calculations required to perform this operation may be too long and / or require too many computing resources to be used on the ship. In this case, it is important to obtain an estimate of the sloshing response as soon as possible and using the shipboard system that may have the lowest cost. Therefore, one idea behind these methods is to pre-perform most of these calculations based on a plurality of input data vectors that can be appropriately selected to cover the entire operating range or functions of the ship, and store the estimated sloshing response of the tank associated with the input data vector in a database for each input data vector among those input data vectors. If the input data vector exists in the database, the estimated sloshing response can be obtained by simply reading the database, otherwise the estimated sloshing response can be obtained in other ways by interpolation based on the database. Compared with the estimation based on the statistical model itself, this requires a significant reduction in calculation time and computing resources. As a result, the statistical model itself does not even need to implement the estimation of the statistical model on the ship, and only the database is sufficient. Then, the database-based estimation can be implemented by a system on the ship or even by a land station communicating with the ship (such as by radio or via satellite).

[0040] According to one embodiment, the present invention also provides a database obtained by the method of obtaining a database as described above.

[0041] According to one embodiment, the present invention also provides a computer-readable storage medium, on which a database obtained by the method of obtaining a database as described above is stored.

[0042] According to an embodiment following a first variant, the present invention also provides a method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, the steps comprising:

[0043] - determining the current filling level of the tank;

[0044] - determining the current sea state;

[0045] - generating an input data vector, the input data vector including the current filling level of the tank and the current sea state determined in this way; and

[0046] - estimating the sloshing response of the tank according to the input data vector generated in this way and according to the database obtained by the method following the first variant.

[0047] According to an embodiment following the second variant, the present invention further provides a method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the steps of:

[0048] - determining the current filling level of the tank;

[0049] - determining the current movement state of the ship;

[0050] - generating an input data vector comprising the current filling level of the tank and the current movement state of the ship determined in this way; and

[0051] - estimating the sloshing response of the tank based on the input data vector generated in this way and based on a database obtained by the method following the second variant.

[0052] Thanks to these methods, the sloshing response of the tank can be estimated by using a statistical model pre-trained on a test database. As described above, this estimation requires much less computing time and computing resources compared to an estimation based on the statistical model itself, and can be implemented by a system on the ship or by a land station communicating with the ship.

[0053] According to one embodiment, multiple tanks are considered and the method includes a previous step of defining the position of each tank on the ship.

[0054] According to one embodiment, the method further comprises the steps of: the steps including providing an alert to the user when the estimated sloshing response of the tank exceeds an alert threshold; and the method further comprises the step of preferably assisting in a decision aimed at reducing sloshing. This decision assistance step may include suggesting changing the direction or course of the ship, changing the heading particularly suitable for a stationary floating structure, changing the speed of the ship or changing the filling level of one or more tanks (between the tanks or between the tanks and a storage facility outside the ship if it is a stationary floating structure).

[0055] Furthermore, the alert may include reporting a problem to be corrected immediately or in the near future, and if possible, the alert designates one or more tanks from among the multiple tanks of the ship that require inspection and maintenance operations for possible repair.

[0056] Thus, a user such as a crew member can take any necessary measures if necessary, such as slowing down or stopping the ship or changing the course of the ship, to limit the sloshing in the tank, thereby reducing the risk of tank damage.

[0057] According to an embodiment following the first variant, the present invention further provides a management system for a ship, the ship including at least one sealed and thermally insulated tank for transporting liquefied gas, the system comprising:

[0058] - At least one filling level sensor for measuring the current filling state of a tank;

[0059] - A device for evaluating the sea state, which can evaluate the current sea state; and

[0060] - A processing device configured to: generate an input data vector including the current filling level of the tank and the current sea state evaluated by the sea state evaluation device; and estimate the sloshing response of the tank based on the input data vector generated in this way and based on a database obtained by following the method of the first variant.

[0061] According to an embodiment following the second variant, the present invention further provides a management system for a ship, the ship including at least one sealed and thermally insulated tank for transporting liquefied gas, the system including:

[0062] - At least one filling level sensor for measuring the current filling state of the tank;

[0063] - A device for evaluating the current moving state of the ship, which can evaluate the current moving state of the ship; and

[0064] - A processing device configured to: generate an input data vector including the current filling level of the tank and the current moving state of the ship; and estimate the sloshing response of the tank based on the input data vector generated in this way and based on a database obtained by following the method of the second variant.

[0065] Such a system achieves the same advantages as the method described above.

[0066] According to one embodiment, the processing device is further configured to provide an alarm to the user when the estimated sloshing response of the tank exceeds an alarm threshold, and preferably is configured to provide assistance to the user in making decisions aimed at reducing sloshing.

[0067] According to another embodiment, the present invention further provides a method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method including the following steps, the steps including:

[0068] - Determining the current filling level of the tank;

[0069] - Estimating the future sea state based on meteorological information and the ship's route;

[0070] - Generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future sea state; and

[0071] - Estimate the future sloshing response of the tank based on the input data vectors generated in this way and based on a database obtained according to a method following the first variant.

[0072] According to the above method, due to a statistical model pre-trained based on test data with the aid of a database, the future sloshing response of the tank can be estimated based on meteorological information and the route of the ship. As mentioned above, compared to an estimate based on the statistical model itself, this estimate requires much less computing time and computing resources and can be implemented by a system on the ship or by a land station communicating with the ship.

[0073] According to another embodiment, the present invention also provides a method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, the steps comprising:

[0074] - Determine the current filling level of the tank;

[0075] - Estimate the future movement state of the ship based on meteorological information and the route of the ship;

[0076] - Generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future movement state of the ship; and

[0077] - Estimate the future sloshing response of the tank based on the input data vectors generated in this way and based on a database obtained according to a method following the second variant.

[0078] According to one embodiment, the method further comprises the following steps: the steps comprising determining the route of the ship and / or changing the filling level of the tank such that the future sloshing response of the tank can be reduced. The expression "route of the ship" means the course of the ship, the speed of the ship or a simple avoidance of a geographical area. For a stationary floating structure (ship, barge), i.e. a floating structure in a fixed position, a change in course is reflected in a change in the angle between the north direction and the longitudinal axis of the structure, so as to orient the floating structure or turn the floating structure around in order to reduce the negative impact of surging and waves on the floating structure in a conventional manner.

[0079] Thus, a user such as a crew member can make a decision to make the ship follow a route that can reduce the future sloshing response of the tank, thereby reducing the risk of tank damage.

[0080] According to another embodiment, the present invention also provides a management system for a ship, the ship including at least one sealed and thermally insulated tank for transporting liquefied gas, the system comprising:

[0081] - At least one filling level sensor for measuring the current filling level of the tank;

[0082] - A sea state estimation device capable of estimating the future sea state based on meteorological information and according to the ship's route; and

[0083] - A processing device configured to: generate input data vectors, each input data vector including the current filling level of the tank and the future sea state estimated by the sea state estimation device; and estimate the future sloshing response of the tank based on the input data vectors generated in this way and according to a database obtained by following the method of the first variant.

[0084] According to another embodiment, the present invention also provides a management system for a ship, the ship including at least one sealed and thermally insulated tank for transporting liquefied gas, the system including:

[0085] - At least one filling level sensor for measuring the current filling level of the tank;

[0086] - A moving state estimation device capable of estimating the future moving state of the ship based on meteorological information and according to the ship's route; and

[0087] - A processing device configured to: generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the future moving state of the ship estimated by the ship moving state estimation device; and estimate the future sloshing response of the tank based on the input data vectors generated in this way and according to a database obtained by following the method of the second variant.

[0088] According to one embodiment, the processing device is further configured to determine the ship's route such that the future sloshing response of the tank can be reduced. Description of the Drawings

[0089] In the following description of specific embodiments of the present invention given in a non-limiting manner with reference only to the drawings, the present invention will be better understood and its other objects, details, features and advantages will become more clearly apparent.

[0090] Figure 1 Figure 1 is a schematic representation of a liquefied gas carrier.

[0091] Figure 2 Figure 2 represents a management system integrated into Figure 1 the ship.​​​​

[0092] Figure 3 Figure 3 represents a management system according to another embodiment.

[0093] Figure 4 Figure 4 is a schematic representation of a test tank sloshing response test device.

[0094] Figure 5 Figure 5 is a flowchart showing a method of obtaining a database that can be used to estimate the sloshing response of a tank.

[0095] Figure 6 Figure 6 is a flowchart showing a method of estimating the sloshing response of a tank.

[0096] Figure 7 Figure 7 is a flowchart showing another method of estimating the sloshing response of a tank.

[0097] Figure 8 Figure 8 is a flowchart showing yet another method of estimating the sloshing response of a tank. DETAILED DESCRIPTION

[0098] The embodiments described below are described with reference to a ship having a double hull forming a support structure in which a plurality of sealed and thermally insulated tanks are arranged. In such a support structure, the tanks have, for example, a polyhedral geometry, such as a prismatic shape.

[0099] Such sealed and thermally insulated tanks are provided, for example, for transporting liquefied gas. The liquefied gas is then transported in such tanks at low temperature, which makes a thermally insulated tank wall necessary to keep the liquefied gas at that temperature. Therefore, it is particularly important to maintain the integrity of the tank wall, including the thermally insulated space located below the sealing membrane, on the one hand to keep the tank sealed and avoid leakage of liquefied gas from the tank, and on the other hand to prevent degradation of the insulation properties of the tank to keep the gas in its liquefied form.

[0100] Such sealed and thermally insulated tanks also include an insulation barrier that is anchored to the double hull of the ship and supports at least one sealing membrane. By way of example, such tanks can be produced according to the technology sold on the market under the applicant's trademark or or other trademarks.

[0101] Figure 1 ​​​​​​​​​​​​There is shown a ship 1 which includes four sealed and thermally insulated tanks 2. The four tanks 2 may have the same or different filling states. When it is at sea, the ship 1 undergoes many movements related to the navigation conditions. These movements of the ship 1 are transmitted to the liquid contained in tanks 3, 4, 5, 6, and the liquid thus undergoes movements in tanks 3, 4, 5, 6. These movements of the liquid in tanks 3, 4, 5, 6 produce impacts on the walls of tanks 3, 4, 5, 6, which, if they are too violent, will immediately damage the tanks. In addition, repeated hammering of the walls of tanks 3, 4, 5, 6 at a high and non-destructive level may cause the said walls to degrade due to wear from fatigue. Now, it is important to maintain the integrity of the walls of tanks 3, 4, 5, 6 in order to maintain the sealing and insulating properties of tanks 3, 4, 5, 6.

[0102] It is well known to avoid critical navigation conditions in order to prevent movements of the liquid that pose a risk of immediate damage to the tanks. However, there is still a need for methods that are able to estimate the sloshing response of the tanks during ship navigation and, if necessary, take the necessary measures to prevent excessive sloshing that poses a risk of damaging the primary sealing membranes of the tanks.

[0103] Figure 2 There is shown an example of a management system 100 on the ship 1. The management system 100 includes a central processing unit 110 which is connected to a plurality of on-board sensors 120 such that various parameter measurements can be obtained. Thus, the sensors 120 include, for example but not exhaustively, at least one filling level sensor 121 for each tank, sensors 122 for various movements of the ship, and a sea condition sensor 123. The management system 100 also includes a communication interface 130 that enables the central processing unit 110 to communicate with remote devices, for example to obtain meteorological data, ship position data, or other data.

[0104] The ship movement sensor 122 determines the measured movements of the ship, for example by measuring the accelerations that the ship undergoes on three perpendicular axes of translation and rotation. In order to evaluate the movements of the ship, an inertial measurement unit (IMU) can be advantageously used, which includes one or more accelerometers and / or one or more gyroscopes such as mechanical gyroscopes, and / or one or more magnetometers. Assuming that a plurality of these (of the same type or two different types) are used, these measurement units are advantageously distributed on the ship, thus producing an accurate measurement of the movements of the ship. It should be noted that an IMU is sometimes commonly referred to as a motion reference unit (MRU).

[0105] The sea condition sensor 123 obtains the current sea conditions in the environment of the ship, such as the height and frequency of the waves in the environment of the ship. For example, in one embodiment, the height and / or frequency of the waves is obtained from the visual observations of the crew.

[0106] The management system 100 further includes a human-machine interface 140. The human-machine interface 140 includes a display device 41. The display device 41 enables an operator to obtain management information calculated by the system or measurements obtained by the sensors 120 or even the current sloshing state, and the current sloshing state can be estimated as described in detail below.

[0107] The human-machine interface 140 further includes a collection device 42. The collection device 42 enables an operator to manually provide values to the central processing unit 110, typically data that cannot be obtained by the sensors because the ship does not include the necessary sensors or the necessary sensors are damaged. For example, in one embodiment, the collection device enables an operator to input information about the height and / or frequency of waves and / or manually input the heading and / or speed of the ship based on visual observation.

[0108] The management system 100 further includes a database 150. The database can be used to estimate the sloshing response of the tank, as will be described in detail below.

[0109] Figure 3 An example of a management system 200 located on land and communicating with a ship 1 is shown. The ship includes a central processing unit 110, sensors 120, and a communication interface 130. The management unit 200 includes a central processing unit 210, a communication interface 230, a human-machine interface 240, and a database 250. The functions of the management system 200 are similar to those of the management system 100, except that the information measured by the sensors 120 on the ship 1 is sent to the management system 200 located on land through the communication interfaces 130 and 230. For example, the communication interface can use terrestrial or satellite radio to transmit data.

[0110] Now, with the help of Figure 4 and Figure 5 it will be described how to obtain the database 150.

[0111] Figure 4 An example of a test device 1000 capable of performing tests on a test tank 1010 is schematically shown. The test includes subjecting the test tank 1010 to movement, where the test tank 1010 has a given filling level of a fluid 1011; and measuring the pressure at at least one point on the wall 1010a of the test tank 1010 and / or the number of impacts on at least one wall of the test tank 1010 using a pressure sensor 1012.

[0112] The test tank 1010 can be smaller than the tank for which the sloshing response is to be estimated and / or have a geometry representative of the tank for which the sloshing response is to be determined.

[0113] Of course, the fluid 1011 preferably has the same properties and ideally has the same temperature, density, and viscosity as the fluid transported in the tank for which the sloshing response is to be determined; it can in particular be liquefied petroleum gas (LPG) at a temperature between, for example, -50°C and 0°C or liquefied natural gas (LNG) at approximately -162°C under atmospheric pressure. A variety of liquefied gases can also be envisaged, in particular methane, ethane, propane, butane, ammonia, dihydrogen, or ethylene.

[0114] Furthermore, the pressure can be measured at several or all of the walls 1010a of the test tank 1010 or even at multiple points on these walls, and the number and arrangement of the pressure sensors 1012 can be adjusted accordingly. If the number of impacts on at least one wall of the test tank 1010 is measured, this measurement is achieved by means of a plurality of pressure sensors 1012 appropriately arranged on this wall. The number of impacts on multiple walls or all the walls of the test tank 1010 can be measured.

[0115] As mentioned above, the test tank 1010 undergoes movement during the test. In the example shown, the device 1000 thus includes a platform 1013 to which the test tank 1010 is fixed. The platform 1013 is driven to move by the action of six hydraulic cylinders 1015, which are connected to the platform at one of their ends at three fixed points 1014 and to the frame or the ground 1001 at the other end. This enables the test tank 1010 to be driven to move with six degrees of freedom in translation and rotation. Of course, the test tank 1010 can be driven to move in different ways.

[0116] The device 1000 also includes a test control unit 1020. The test control unit 1020 is configured to control the hydraulic cylinders 1015 so that the test tank 1010 undergoes a predetermined movement in the test program. In one embodiment, these movements are movements representative of a given movement of a ship, which preferably takes into account the position of the tank on the ship and / or the geometry of the tank. In another embodiment, these movements are movements representative of a given sea state, which are converted into the corresponding movement of the ship, preferably taking into account the position of the tank on the ship and / or the geometry of the tank. Evaluating the corresponding movement of the ship based on a given sea state is a conventional task for evaluating the seaworthiness of the ship. In addition, the test control unit 1020 stores the values measured by at least one pressure sensor 1012 during the test.

[0117] The test control unit 1020 communicates with the test data processing unit 1030. The test data processing unit 1030 includes a communication interface 1031, which is capable of receiving from the test control unit 1020 the values measured by at least one pressure sensor 1012 during a test and the movement applied to the test tank 1010 during the test. The test data processing unit 1030 also includes a memory 1033 and a central processing unit 1032.

[0118] The test data processor unit 1030 is configured to train a statistical model in the central processor 1032 communicating with the memory 1033 by a machine learning method. The statistical model is capable of estimating the sloshing response of the tank based on the filling level of the tank and the current sea state and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship. The sloshing response includes at least one of the following: the number of impacts of the fluid on the tank wall, the maximum pressure on the tank wall, and the likelihood of damaging the tank. In a variant, the statistical model considers multiple tanks, and the statistical model is capable of estimating the sloshing response of each tank based on the position of each tank on the ship.

[0119] More specifically, the statistical model is trained by a supervised machine learning method. For example, the supervised machine learning method can be a Gaussian process regression method. Gaussian process regression methods are well known in themselves; they are very suitable for training statistical models because they are able to produce statistical models that can solve regression problems for any input data set based on training on a relatively limited amount of data. Nevertheless, other supervised machine learning methods can also be employed.

[0120] The statistical model is trained based on the test results generated using the test tank 1010. More specifically, in a preferred example, the statistical model is trained based on the sloshing response of the test tank 1010 during each test, which is pre-calculated based on the values measured by at least one pressure sensor 1012 during the test. The sloshing response of the test tank 1010 can include at least one of the following: the number of fluid impacts on one or some of the walls 1010a of the test tank 1010 within a given time period and the maximum pressure on the walls 1010a of the test tank 1010. In a variant, the statistical model is trained based on the results of the tests performed on the test tank 1010 and based on test data obtained or measured on one or more tanks that act as test tanks 1010 on ships that are liquefied gas carriers and / or user services. In another variant, the statistical model can be trained only based on test data obtained or measured on one or more tanks that act as test tanks 1010 on ships that are liquefied gas carriers and / or user services.

[0121] Now with the help of Figure 5A method 300 for enabling access to a database 150 will be described. Steps 301 to 305 may be performed by a central processing unit 1032 communicating with a memory 1033.

[0122] Method 300 may optionally include step 301, which includes excluding from a test dataset used for training a statistical model any test results that show the sloshing response of test tank 1010 to be below a certain threshold. As a result, the statistical model is trained only on test data of significant sloshing that has been revealed in test tank 1010, which improves the accuracy of the estimation of the sloshing response using the statistical model.

[0123] After optional step 301, method 300 includes step 302, which includes training the statistical model as described above.

[0124] During step 302, at least one constraint may optionally be imposed on the statistical model during its training by a supervised machine learning method. These constraints may be defined based on fundamental physical considerations such as no sloshing when the fill level of the tank is zero and / or on considerations obtained through practical experience such as the fact that the greater the movement of the tank or the greater its size, the higher the potential sloshing response. The result is an improvement in the accuracy of the estimation of the sloshing response by the statistical model.

[0125] After completion of step 302, a statistical model is obtained that is capable of estimating the sloshing response of the tank based on the fill level of the tank, the current sea state, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship, for any value of these quantities, including those not tested on test tank 1010. However, the calculations required to perform this operation may be too long and / or require too much computational resources to implement on a ship, so it is important to obtain an estimate of the sloshing response as soon as possible and using the shipboard system that may be the least costly. This is the reason for using step 303 after step 302, which includes generating a plurality of input data vectors, each input data vector including the fill level of the tank and the current sea state, followed by step 304, which includes for each input data vector generated in step 303: obtaining the estimated sloshing response of the tank by means of the statistical model in step 302 and storing the estimated sloshing response associated with the input data vector in a database.

[0126] In step 305, the database obtained in step 304 is optionally transmitted to a management system 100 or stored on a computer-readable storage medium. A database 150 is also obtained, which will be described below.

[0127] So far, a situation has been described in which a statistical model is capable of estimating the sloshing response of a tank based on at least the filling level of the tank and the current sea state. However, in a variant, the statistical model is capable of estimating the sloshing response of the tank based on the filling level of the tank, the current movement state of the ship, and at least one of the following optional items: the draft of the ship, the speed of the ship, and the course of the ship. Then steps 302, 303, 304 are modified accordingly.

[0128] Now, a method 400 for estimating the sloshing response of a tank with the aid of a database 150 will be described with the aid of Figure 6 Method 400 can consider multiple tanks rather than just one tank. In this case, before performing method 400, a preliminary step of defining the position of each tank in the ship can be performed.

[0129] According to a first embodiment, Figure 6 The flowchart of is executed entirely in the central processing unit 110 forming a single processing device. According to a second embodiment, Figure 6 The flowchart of is partially executed in a land management system 200 communicating with the ship. According to this second embodiment, the ship 1 transmits all information from the sensors 120 to the land station, and the central processing unit 110 and the central processing unit 210 together form a shared processing device.

[0130] Method 400 includes a first step 401, and the first step 401 includes determining the current filling level and the current sea state of the tank. The current filling level of the tank is typically determined based on the filling indication provided by the filling level of the tank sensor 121. The current sea state can be determined based on the indication provided by the sea condition sensor 123 and / or through a land or satellite radio communicating with a network of weather stations.

[0131] In step 401, optionally, the draft and / or the course of the ship are also determined, typically based on the indication provided by the shipborne system of the ship. The draft of the ship is typically provided to the shipborne system of the ship by one or more floating and / or hydrostatic pressure type sensors. The course of the ship is typically provided to the shipborne system of the ship by one or more navigation compasses.

[0132] Method 400 further includes a second step 402, and the second step 402 includes generating an input data vector including the data determined in step 401.

[0133] The method 400 further comprises a third step 403, which comprises estimating the slosh response of the tank based on the database 150 and the input data vector generated in step 402. More specifically, if the input data vector is proven to be present in the database 150, the slosh response may be obtained by simply reading the database 150. However, the database 150 will typically not contain the input data vector, but rather contain input data close to that contained in the input data vector. In this case, the slosh response will be obtained by interpolating from the slosh responses associated with two or more adjacent input data vectors present in the database 150.

[0134] After step 403, the obtained sloshing response may be compared to an alarm threshold and if the sloshing response exceeds the alarm threshold, an alarm may be displayed to a user, for example on the display device 41. The display of the alarm is preferably followed by a step of assisting in making a decision aimed at reducing sloshing. The decision-assisting step may include suggesting a change in the direction or course of the vessel, a change in heading particularly applicable to a stationary floating structure, a change in the speed of the vessel, or a change in the fill level of one or more tanks (between tanks or, in the case of a stationary floating structure, between a tank and a storage facility external to the vessel). In addition, the alarm may include reporting a problem to be corrected immediately or in the short term, if possible specifying one or more tanks requiring inspection and maintenance operations for possible repair.

[0135] Now we will use Figure 7 Another method 500 for estimating the sloshing response of a tank with the aid of a database 150 is described. In this variant, the database 150 is obtained from a statistical model capable of estimating the sloshing response of the tank based on the filling level of the tank, the current movement state of the vessel and optionally at least one of the following: the draft of the vessel, the speed of the vessel and the heading of the vessel, as described above with reference to Figure 4 The method 500 may take into account a plurality of tanks rather than just one tank. In this case, before executing the method 500, there may be a previous step to define the position of each of the tanks on the vessel.

[0136] The method 500 comprises a first step 501 which comprises determining a current filling level of the tank and a current movement state of the vessel. The current filling level of the tank is typically determined based on a filling indication provided by a filling level of the tank sensor 121. The current movement state of the vessel may be determined based on an indication provided by a vessel movement sensor 122.

[0137] In step 501, optionally, the draft of the ship and / or the heading of the ship are also determined, typically based on indications provided by the shipborne systems of the ship. The draft of the ship is typically provided to the shipborne systems of the ship by one or more floating and / or hydrostatic pressure type sensors. The heading of the ship is typically provided to the shipborne systems of the ship by one or more navigation compasses.

[0138] The ship motion sensor 122 typically has a much higher acquisition frequency than the typical duration of the sloshing evolution of the tank, and the indications provided by the ship motion sensor 122 can be averaged during the acquisition period, and then other data determined in step 501 can be averaged during the same acquisition period.

[0139] Method 500 further includes a second step 502 similar to step 402, and the second step 502 includes generating an input data vector including the data determined in step 501.

[0140] Method 500 further includes a third step 503, and the third step 503 includes estimating the sloshing response of the tank based on the database 150 and the input data vector generated in step 502. Step 503 is similar to step 403, so it will not be explained in detail here.

[0141] After step 503, the obtained sloshing response can be compared with an alarm threshold, and if the sloshing response exceeds the alarm threshold, an alarm can be displayed to the user, for example, on the display device 41. After displaying the alarm, steps for assisting in making decisions aimed at reducing sloshing can preferably follow. These decision-making assistance steps can include suggesting changing the direction or course of the ship, changing the heading that is particularly applicable to a stationary floating structure, changing the speed of the ship, or changing the filling level of one or more tanks (between the tanks or, if it is a stationary floating structure, between the tanks and storage facilities outside the ship). In addition, the alarm can include reporting problems to be corrected immediately or in the short term, and if possible, the alarm specifies one or more tanks that require inspection and maintenance operations for possible repair.

[0142] Now with the aid of Figure 8 Another method 600 for estimating the sloshing response of the tank with the aid of the database 150 will be described. In this variant, the database 150 is obtained from a statistical model that is capable of estimating the sloshing response of the tank based on the filling level of the tank, the current sea state, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship.

[0143] Method 600 includes a first step 601 that includes determining a current fill level of a tank and estimating future sea conditions. The current fill level of the tank is typically determined based on a fill indication provided by a tank fill level sensor 121. The future sea conditions are estimated based on meteorological information and the course of the ship. The course of the ship is typically obtained from indications provided by the ship's on-board systems such as the ship's speed and the ship's heading. The meteorological information may be provided by a sea condition sensor 123 and / or by a terrestrial or satellite radio communicating with a network of weather stations.

[0144] Method 600 further includes a second step 602 that includes generating a plurality of input data vectors, each input data vector including the current fill level of the tank and the estimated future sea conditions.

[0145] In step 601, optionally, the draft of the ship, the heading of the ship, and the speed of the ship are also determined, typically according to indications provided by the ship's on-board systems. The draft of the ship is typically provided to the ship's on-board systems by one or more floating and / or hydrostatic pressure type sensors. The heading of the ship is typically provided to the ship's on-board systems by one or more navigation compasses. The speed of the ship is typically provided to the ship's on-board systems by an IMU and / or a GPS type satellite navigation receiver.

[0146] Method 600 further includes a third step 603 that includes estimating a future sloshing response of the tank according to a database 150 and each input data vector generated in step 602. Step 603 is similar to step 403 and will not be explained in detail here.

[0147] After step 603, the course of the ship can be determined such that the future sloshing response of the tank can be reduced relative to the sloshing response of the ship that would occur if the ship maintained its current course. The so-called expression "course of the ship" means the heading of the ship, the speed of the ship, or a simple avoidance of a geographical area. For a stationary floating structure (ship, barge), i.e., a structure with a fixed position, a change in azimuth is a change in the angle between the north direction and the longitudinal axis of the structure, so as to orient the floating structure or turn around the floating structure in a classical way to reduce the negative impact of surging and waves on the floating structure. Additionally or alternatively, a change in the fill level of the tank can be determined such that the future sloshing response of the tank can be reduced.

[0148] A variant of method 600 from Figure 8 is described below. In this variant, database 150 is obtained from a statistical model that is capable of estimating the sloshing response of the tank according to the fill level of the tank, the current movement state of the ship, and optionally at least one of the following: the draft of the ship, the speed of the ship, and the heading of the ship.

[0149] In this variant, step 601 includes determining the current filling level of the tank and estimating the future movement state of the ship. The current filling level of the tank is typically determined based on the filling indication provided by the tank filling level sensor 121. The future movement state of the ship is estimated based on meteorological information and the ship's route. The ship's route is typically obtained from indications provided by the ship's on-board systems such as the ship's speed and the ship's heading. The meteorological information can be provided by the sea condition sensor 123 and / or by a terrestrial or satellite radio communicating with a network of weather stations. According to one example, the future movement state of the ship can be estimated as follows: initially estimating the future sea conditions based on the meteorological information and the ship's route, and then in a second step estimating the future movement state of the ship based on the future sea conditions estimated in this way. Remember, as mentioned above, assessing the corresponding movement of the ship based on a given sea condition is a routine task for assessing the seaworthiness of the ship.

[0150] Step 602 then includes generating a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future movement state of the ship.

[0151] In step 601, it is also optionally determined the draft of the ship, the heading of the ship, and the speed of the ship, as mentioned above.

[0152] Step 603 includes estimating the future sloshing response of the tank according to each input data vector generated in step 602 and according to the database 150. Step 603 is similar to step 403, and thus will not be explained in detail.

[0153] After step 603, the ship's route can be determined such that the future sloshing response of the tank can be reduced relative to the sloshing response of the tank that would occur if the ship maintained its current route, as already mentioned above.

[0154] Although the present invention has been described in connection with multiple specific embodiments, it is obvious that the present invention is in no way limited to these specific embodiments, and if these specific embodiments fall within the scope of the present invention, the present invention includes all technical equivalents of the described devices and their combinations.

[0155] In addition, it is obvious that the features or feature combinations described with reference to a method are equally applicable to the corresponding system, and vice versa.

[0156] The use of the verbs "comprise", "include" and their conjugate forms does not exclude the presence of elements or steps other than those listed in the claims.

[0157] In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim.

Claims

1. A method for obtaining a database (150) that can be used to estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas, the method comprising the following steps, The steps include: Training a statistical model based on a test data set by a supervised machine learning method, the statistical model being capable of estimating the sloshing response of the tank based on the filling level of the tank and the current sea state, and the test data set being obtained from the results of a plurality of tests, each test including: subjecting a test tank (1010) having a given filling level to movement; and measuring the pressure at at least one point on the wall (1010a) of the test tank (1010) and / or the number of impacts on at least one wall of the test tank (1010); Generating a plurality of input data vectors, each input data vector including the filling level of the tank and the current sea state; and For each input data vector generated in this way: obtaining the estimated sloshing response of the tank by means of the statistical model; and storing the estimated sloshing response of the tank associated with the input data vector in a database.

2. The method according to claim 1, wherein, The sloshing response includes at least one of the following items: the number of impacts of the fluid on the wall of the tank, the maximum pressure on the wall of the tank, and the likelihood of damaging the tank.

3. The method according to claim 1, wherein, Applying at least one constraint to the statistical model during the training of the statistical model by the supervised machine learning method.

4. The method according to claim 1, wherein, The method further includes the step of: before the step of training the statistical model based on the test data set by the supervised machine learning method, excluding test results characterized by a sloshing response below a threshold from the test data set.

5. The method according to claim 1, wherein, The statistical model is capable of estimating the sloshing response of the tank based on the filling level of the tank and the current sea state and at least one of the following items: the draft of the ship, the speed of the ship, and the heading of the ship.

6. A method for obtaining a database (150) that can be used to estimate the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas, the method Including the following steps, the steps including: Training a statistical model based on a test data set by a supervised machine learning method, the statistical model being capable of estimating the sloshing response of the tank based on the filling level of the tank and the current movement state of the ship, and the test data set being obtained from the results of a plurality of tests, each test including: subjecting a test tank (1010) having a given filling level to movement; and measuring the pressure at at least one point on the wall (1010a) of the test tank (1010) and / or the number of impacts on at least one wall of the test tank (1010); Generating a plurality of input data vectors, each input data vector including the filling level of the tank and the current movement state of the ship; and For each input data vector generated in this way: obtaining the estimated sloshing response of the tank by means of the statistical model; and storing the estimated sloshing response of the tank associated with the input data vector in a database.

7. The method according to claim 6, wherein, The sloshing response includes at least one of the following items: the number of impacts of the fluid on the wall of the tank, the maximum pressure on the wall of the tank, and the likelihood of damaging the tank.

8. The method according to claim 6, wherein, Applying at least one constraint to the statistical model during the training of the statistical model by the supervised machine learning method.

9. The method according to claim 6, wherein, The method further includes the following steps: before the step of training the statistical model based on a test data set by means of supervised machine learning, excluding from the test data set test results characterized by a sloshing response below a threshold.

10. A method for estimating the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, the steps comprising: Determine the current filling level of the tank; Determine the current sea state; Generate an input data vector including the current filling level of the tank and the current sea state determined in this way; And Based on the input data vector generated in this way and Estimate the sloshing response of the tank based on the database (150) obtained by the method according to claim 1.

11. A method for estimating the sloshing response of at least one sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, the steps comprising: Determine the current filling level of the tank; Determine the current movement state of the ship; Generate an input data vector including the current filling level of the tank and the current movement state of the ship determined in this way; and Estimate the sloshing response of the tank based on the input data vector generated in this way and based on the database (150) obtained by the method according to claim 6.

12. The method according to claim 10 or 11, wherein, Multiple tanks are considered and the method includes a previous step of defining the position of each of the tanks in the ship.

13. The method according to claim 10 or 11, wherein, The method further includes the following steps: Providing an alert to a user when the estimated sloshing response of the tank exceeds an alert threshold.

14. The method according to claim 13, wherein, The method further includes: a step of assisting in making decisions aimed at reducing sloshing.

15. A management system for a ship, the ship including at least one sealed and thermally insulated tank (2) for transporting liquefied gas, the management system comprising: At least one filling level sensor (121) for measuring the current filling state of the tank (2); A sea state assessment device (123) for assessing the sea state, the sea state assessment device (123) being capable of assessing the current sea state; and A processing device (110) configured to: generate an input data vector including the current filling level of the tank and the current sea state evaluated by the sea state assessment device (123); and estimate the sloshing response of the tank (2) based on the input data vector generated in this way and based on the database (150) obtained by the method according to any one of claims 1 to 5.

16. A management system for a ship, the ship (1) including at least one sealed and thermally insulated tank (2) for transporting liquefied gas, the management system comprising: At least one filling level sensor (121) for measuring the current filling state of the tank (2); A device (122) for evaluating the current movement state of the ship, the device (122) for evaluating the current movement state of the ship being capable of evaluating the current movement state of the ship; and A processing device (110) configured to: generate an input data vector including the current filling level of the tank and the current movement state of the ship; and estimate the sloshing response of the tank based on the input data vector generated in this way and based on the database (150) obtained by the method according to any one of claims 6 to 9.

17. A method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, The steps include: Determine the current filling level of the tank; Estimate future sea conditions based on meteorological information and the route of the ship; Generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future sea conditions; and Estimate the future sloshing response of the tank based on the input data vectors generated in this way and based on the database (150) obtained by the method according to claim 1.

18. A method for estimating the sloshing response of a sealed and thermally insulated tank for transporting liquefied gas on a ship, the method comprising the following steps, The steps include: Determine the current filling level of the tank; Estimate the future movement state of the ship based on meteorological information and the route of the ship; Generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future movement state of the ship; and Estimate the future sloshing response of the tank based on the input data vectors generated in this way and based on the database (150) obtained by the method according to claim 6.

19. The method according to claim 17 or 18, further comprising the following steps: Determine the route of the ship and / or change the filling level of the tank so as to reduce the future sloshing response of the tank.

20. A management system for a ship, the ship (1) including at least one sealed and thermally insulated tank (2) for transporting liquefied gas, the management system comprising: At least one filling level sensor (121), the at least one filling level sensor (121) for measuring the current filling level of the tank (2); A sea condition estimation device (123), the sea condition estimation device (123) being capable of estimating future sea conditions based on meteorological information and based on the route of the ship (1); and A processing device (110), the processing device (110) being configured to: generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the future sea conditions estimated by the sea condition estimation device (123); and based on the input data vectors generated in this way and Estimate the future sloshing response of the tank based on the database (150) obtained by the method according to claim 1.

21. A management system for a ship, wherein the ship (1) includes at least one sealed and thermally insulated tank (2) for transporting liquefied gas, and the management system includes: At least one filling level sensor (121), the at least one filling level sensor (121) for measuring the current filling level of the tank (2); A movement state estimation device, the movement state estimation device being capable of estimating future sea conditions based on meteorological information and based on the route of the ship; and A processing device (110), the processing device (110) being configured to: generate a plurality of input data vectors, each input data vector including the current filling level of the tank and the estimated future movement state of the ship by the ship movement state estimation device; and based on the input data vectors generated in this way and based on the database (150) obtained by the method according to claim 6, estimate the future sloshing response of the tank.

22. The management system according to claim 20 or 21, wherein The processing device (110) is further configured to: determine the route of the ship so as to reduce the future sloshing response of the tank.

Images