Method and system for monitoring a marine structure

Abstract

Method and system for monitoring a marine structure The monitoring method carries out the monitoring of a marine structure using instrumentation comprising sensors positioned on the marine structure (2) at determined measurement points, the monitoring method comprising in situ and online monitoring of the marine structure (2) implemented by an electronic monitoring device (24) by assimilation of the measurements provided by the sensors (10) using a reduced order model (12) configured for the calculation of a current state (CS) of the marine structure (2) from the measurements provided by the sensors (10), the current state (CS) of the marine structure (2) including estimation points distinct from the measurement points where the sensors (10) are located. Figure 1

Description

Title of the invention: Method and system for monitoring a marine structure

[0001] The present invention relates to the field of monitoring a marine structure, such as a structure of a ship or a structure of a mobile or stationary marine platform.

[0002] A marine structure is subject to significant stresses that are difficult to predict in advance. It is desirable to monitor marine structures to understand the evolution of their state of health, for example to prevent any failure and / or to ensure the maintenance of the marine structure.

[0003] It is possible to carry out monitoring of a marine structure using data from instrumentation comprising a plurality of sensors attached to the marine structure.

[0004] The instrumentation cannot, however, cover the entire marine structure and critical areas presenting, for example, strong stress gradients are sometimes difficult to access and devoid of sensors such as deformation sensors or stress sensors.

[0005] In addition, measurement uncertainties, linked for example to the intrinsic presence of noise, to the lack of knowledge of environmental variables and to the uncertainty regarding the exact position of the sensors, can significantly reduce the precision of the measurements and therefore complicate the interpretation of the measurements.

[0006] Thus, establishing monitoring of marine structures solely from the measurement signals of such instrumentation is not satisfactory.

[0007] It is also possible to monitor a marine structure by relying solely on measurement signals from sensors placed at predetermined weak points where the failure mode is known a priori.

[0008] However, this is not applicable for a marine structure for which the failure modes and their location are not known a priori or may vary from one marine structure to another, as is the case for marine structures produced individually or in small series.

[0009] The increase in processing capacities of big data (or "Big Data") and deep learning techniques allow us to envisage the possibility of producing and operating instrumentation of a marine structure with a very large number of sensors and a fine mesh of the marine structure.

[0010] However, such a solution would be prohibitively expensive both in terms of instrumentation, data storage and data processing, and would be difficult to implement for ships, particularly for submarines, due to the delicate navalization of the instrumentation, notably with the need to multiply the hull passages for the installation of sensors and their connections to a measurement processing and analysis system.

[0011] Furthermore, the availability of big data in sufficient quantity for the implementation of deep learning is delicate in the marine domain for which certain marine structures are produced in very small series.

[0012] It is possible to associate data from sensors placed on the marine structure with a structural model (“model-based structural health monitoring” in English) to obtain a global estimate of the deformations and constraints of the marine structure from the locally measured data.

[0013] For this, it is possible to use a finite element structural model capable of estimating deformations and stresses as a function of a given loading. However, such a finite element structural model is very heavy and expensive in terms of calculation time, which makes its use in situ and online very difficult.

[0014] It is possible to estimate the fatigue of a marine structure by using measurements provided by sensors positioned on the marine structure, radar measurements of the waves to which the marine structure is subjected in order to determine the stresses on the marine structure, and a finite element structural model of the marine structure. WO2016 / 108183A1 discloses a system for monitoring the marine structure of a ship configured to carry out such an estimation of the fatigue of a structure of a ship.

[0015] One of the aims of the invention is to propose a method for monitoring a marine structure which is simple to implement and which allows monitoring to be carried out in situ and online, i.e. during operation of the marine structure.

[0016] To this end, the invention proposes a method for monitoring a marine structure using instrumentation comprising sensors positioned on the marine structure at determined measurement points, the monitoring method comprising in situ and online monitoring of the marine structure implemented by an electronic monitoring device by assimilation of the measurements provided by the sensors using a reduced order model configured for calculating a current state of the marine structure from the measurements provided by the sensors, the current state of the marine structure including estimation points distinct from the measurement points where the sensors are located.

[0017] The use of a reduced-order model of the marine structure suitable for the assimilation of measurements, i.e. suitable for providing a current state of the marine structure (e.g. deformation and stress) for a large set of points for estimating the marine structure from measurements carried out on a small set of measurement points to which the sensors are placed, allows the current state of the marine structure to be estimated quickly and with reasonable processing capacities compatible with in situ and online monitoring.

[0018] According to particular implementation modes, the monitoring method comprises one or more of the following characteristics, taken individually or according to all technically possible combinations:

[0019] - the assimilation is carried out by an empirical interpolation method;

[0020] - the assimilation is carried out by a low-data formulation method background setting;

[0021] - assimilation is carried out by a hybrid twin method;

[0022] - each sensor is a strain sensor or a strain sensor;

[0023] - the current state calculated using the reduced order model includes the deformation and / or the constraint at said marine structure estimation points;

[0024] - the monitoring method comprises the calculation, by the electronic device of monitoring, status indicators based on the current status of the marine structure, and transmission of status indicators to a monitoring team;

[0025] - monitoring indicators include an estimate of the lifespan re of the marine structure, an alert in the event of over-stressing of the marine structure an alert in the event of abnormal behavior of the marine structure, a location of possible damage to the marine structure, the location of possible damage being possibly associated with a level of criticality;

[0026] - the marine structure belongs to a surface ship, a submarine ship or a marine platform;

[0027] - the monitoring method comprises:

[0028] - obtaining a higher order structural model representative of the structure marine, the structural model being capable of receiving as input parameters and a loading of the marine structure and of providing as output the deformations and / or the stresses of the marine structure calculated as a function of the parameters and the loading received as input; and

[0029] - the calculation of the reduced order model and the locations for the sensors from the structural model, of a set of constraints;

[0030] - the reduced order model is calculated by an empirical interpolation method;

[0031] - the reduced order model is calculated by an orthogonal decomposition method to eigenvalues.

[0032] The invention also relates to a system for monitoring a marine structure, comprising a set of sensors positioned on the marine structure at measurement points and an electronic monitoring device configured for implementing a method for monitoring the marine structure as defined above.

[0033] The invention also relates to a computer program product recordable on a memory and executable by a processor, the computer program product being configured for the implementation of a method for monitoring a marine structure as defined above.

[0034] The invention and its advantages will be better understood on reading the description which follows, given solely as a non-limiting example, and made with reference to the appended drawings, in which:

[0035] - [Fig.l] [Fig.l] is a schematic view of a marine structure equipped with a marine structure monitoring system;

[0036] - [Fig.2] [Fig.2] is a diagram illustrating the determination of an order model reduced marine structure and associated sensor positions on the marine structure.

[0037] With reference to [Fig.l], a marine structure 2 is equipped with a monitoring system 4 configured for monitoring the mechanical health of the marine structure 2 in situ and online.

[0038] By in situ and online, it is understood that the monitoring system 4 is configured to monitor the marine structure 2 during the operation of the marine structure 2 by analyzing measurements taken on the marine structure 2 during its operation.

[0039] The marine structure 2 is for example floating, and in particular suitable for navigation. The marine structure 2 is for example the structure of a ship 6. The ship 6 is for example a surface ship or a submarine ship. Alternatively, the marine structure 2 is stationary. It is for example a fixed or floating marine platform.

[0040] The monitoring system 4 has instrumentation 8 comprising a plurality of sensors 10. Each sensor 10 is positioned on the marine structure 2 at a measurement location or measurement point and configured to measure a physical quantity of the marine structure.

[0041] In particular, each sensor 10 is configured to measure a deformation of the marine structure 2 at the measurement point where the sensor 10 is placed and / or a stress of the marine structure 2 at the measurement point where the sensor 10 is placed.

[0042] Each sensor 10 is for example a strain gauge configured to measure a deformation of the marine structure 2 at the measurement point where the sensor 10 is placed or a strain gauge configured to measure a strain of the marine structure 2 at the measurement point where the sensor 10 is placed.

[0043] The monitoring system 4 comprises a reduced order model 12 specific to assimilation, i.e. a reduced order model configured for the estimation of a current state CS of the marine structure 2 including the estimation of the state of the marine structure at estimation points distinct from the measurement points at which the the sensors 10, from the sole knowledge of the MC measurements provided by the sensors 10 arranged at the measurement points.

[0044] The reduced order model 12 is a mathematical model (or numerical model) representative of the mechanical behavior of the marine structure 2 and making it possible to calculate physical quantities of the marine structure 2 from MC measurements provided by the sensors 10.

[0045] A structural model of the marine structure 2, such as a finite element model, models the marine structure 2 in detail and makes it possible to determine a state of the marine structure as a function of parameters, such as material parameters, boundary conditions and environmental variables, and of a loading of the marine structure 2, i.e. external forces experienced by the marine structure 2.

[0046] The reduced order model 12 specific to assimilation is distinguished from such a structural model in that it is simplified to take into account a reduced number of input data, here measurements of physical quantities measured on the marine structure 2, such as deformations and / or stresses.

[0047] Preferably, the reduced order model 12 specific to the assimilation is configured to receive as input measurements of physical quantities of the marine structure 2 carried out at measurement points, and to allow the calculation of an estimate of these physical quantities at estimation points distinct from the measurement points, the estimation points being in greater number than the measurement points.

[0048] The monitoring system 6 comprises a data-model hybridization module 14 configured to carry out an assimilation of the MC measurements provided by the sensors 10 by the reduced order model 12 and thus estimate the current state CS of the marine structure 2 as a function of the reduced order model 12 and the measurements provided by the sensors 10.

[0049] In an exemplary embodiment, the data-model hybridization module 14 is configured to use the reduced order model 12 to estimate the current state CS of the marine structure 2 on all of the estimation points as a function of the MC measurements provided by the sensors 10 located at the measurement points.

[0050] To do this, the data-model hybridization module 14 is for example configured to carry out the assimilation for an empirical interpolation method (or EIM for “Empirical Interpolation Method”).

[0051] In an exemplary embodiment, the data-model hybridization module 14 is configured to update the reduced order model 12 to estimate the current state CS of the marine structure 2 on all of the estimation points as a function of the measurements MC carried out by the sensors 10 at the measurement points and in addition to update the reduced order model 12 as a function of the measurements carried out by the sensors 10 at the measurement points.

[0052] In other words, the data-model hybridization module 14 is in this case configured to implement learning to use the MC measurements provided by the sensors 10 for updating the reduced order model 12.

[0053] To do this, the data-model hybridization module 14 is for example configured to carry out the assimilation by a parameterized background data weak formulation (or “Parametrized Background Data Weak Formulation”).

[0054] With this approach, the measured data make it possible on the one hand to calibrate the reduced-order model and on the other hand to learn a correction (mathematically defined as the smallest correction on a reduced-dimension basis making it possible to minimize the difference between the predictions of the state of the system and the locally measured data). The particularities and adaptations to the present case concern: potentially different quantities for the measurement (example in deformation) and the prediction of the state (example in stress), measurable zones potentially different from the zones of interest for the estimation of the state, and the definition of a reduced correction space oriented by the physics involved.

[0055] In yet another exemplary embodiment, the data-model hybridization module 14 is configured to perform the assimilation by a hybrid twin method (or “Hybrid Twin”).

[0056] Learning in this case also concerns the corrective part of the model, obtained by online modification of the structural matrices of the problem under parsimony constraints, via the exploitation of locally measured data. This approach makes it possible to obtain a more precise estimate of the corrected state, as well as the location of any damage.

[0057] Such data-model hybridization methods 14 each allow the implementation of learning.

[0058] The current state CS of the marine structure 2 estimated by the data-model hybridization module 14 comprises for example an estimation of the deformation of the marine structure 2 and / or an estimation of the stress of the marine structure 2 at estimation points including points of the marine structure 2 distinct from the measurement points where the sensors 10 are arranged.

[0059] In a particular embodiment, the sensors 10 are deformation sensors, the assimilation of the measurements for the reduced order model 12 making it possible to calculate a current state CS estimating the deformation and / or the stress at a plurality of estimation points distinct from the measurement points.

[0060] The reduced order model 12 makes it possible to estimate the deformation and / or the stress in the marine structure 12 at a very large number of estimation points, from a small number of measurement points.

[0061] The monitoring system 6 optionally comprises a processing and analysis module 16 configured for processing and analyzing the current state CS estimated at from the measurements provided by the sensors 10 and the reduced order model 12.

[0062] The processing and analysis module 16 is for example configured for the calculation of monitoring indicators from the current state CS of the marine structure 2 and the transmission of the monitoring indicators to a monitoring team, comprising for example at least one human operator.

[0063] The transmission of the status indicators comprises in particular the transmission of the status indicators to a display system 18. The display system 18 comprises for example a dashboard and / or one or more display screens (not shown).

[0064] The monitoring indicators include an estimate of the residual lifespan of the marine structure 2, an alert in the event of over-stressing of the marine structure 2, an alert in the event of abnormal behavior of the marine structure 2, a location of possible damage to the marine structure 2, the location of possible damage being possibly associated with a level of criticality.

[0065] A monitoring method comprises a first phase - or "offline" phase - of calculating the positions of the measurement points of the marine structure 2 at which the sensors 10 must be positioned and of calculating the reduced order model 12, and a second phase - or "online" phase - of exploiting the reduced order model 12 for in situ and online monitoring of the marine structure 2.

[0066] As illustrated in [Fig.2], the offline phase comprises a first step of obtaining a structural model 20 of the marine structure 2.

[0067] The structural model 20 of the marine structure 2 is for example a mathematical model (or digital model) making it possible to calculate a state of the marine structure 2 from parameters and a loading of the marine structure 2.

[0068] The parameters are for example material parameters, boundary conditions, environmental variables and geometric parameters.

[0069] The loading of the marine structure 2 comprises all of the forces to which the marine structure 2 is subjected.

[0070] In the case of a marine structure 2, the latter is in particular subjected to forces generated by the expanse of water on which the marine structure 2 is located, and in particular by waves propagating on this expanse of water.

[0071] The structural model 20 is for example a finite element model.

[0072] For a marine structure 2, generally complex and large, such a model structural 20 is often very expensive to produce and very expensive to use, and in particular requires significant computing power and / or significant computing time to run the structural model 20.

[0073] Preferably, the structural model 20 is capable of receiving as input parameters and a loading of the marine structure 2 and of providing as output the deformations and / or the stresses of the marine structure 2 calculated as a function of the parameters and the loading received entry.

[0074] The structural model 20 is a higher order model than the reduced order model 12. The structural order model 20 takes into account a number N of degrees of freedom strictly greater than the number M of variables taken into account by the reduced order model 12.

[0075] Furthermore, the structural model 20 is not suitable for data assimilation because it takes as input the unknown or uncontrolled parameters online. The reduced order model produced is suitable for assimilation, because the unknown or uncontrolled parameters do not have to be entered, only the local measurements are necessary.

[0076] The first phase comprises a second step of calculating the reduced order model 12 and the locations for the sensors 10 from the structural model 20.

[0077] Advantageously, the calculation of the reduced order model 12 and the locations for the sensors 10 from the structural model 20 takes into account a set of CF constraints, including for example a set of instrumentation constraints indicating the constraints linked to the equipment of the marine structure 2 using the sensors and / or a set of measurable field constraints indicating measurable stress fields and stress fields of interest.

[0078] In an exemplary embodiment, the reduced order model 12 and the measurement points of the sensors 10 are calculated from the structural model 20 by an empirical interpolation method. A database is firstly constituted by grouping the results (in deformations and / or stresses) of the structural model 20 for numerous values ​​of the parameters (of the order of a few hundred). These values ​​are for example randomly drawn from the space of possible parameters. Then, the reduced spatial base is generated iteratively, by searching at iterate k for the parameter which maximizes the difference between the estimation of the state (with the interpolant of rank k-1) and the reference results. This makes it possible to obtain the new vector of the base at iterate k. Then the point at which to position the sensor 10 is the point maximizing this new vector.

[0079] In an exemplary embodiment, the reduced-order model 12 and the measurement points of the sensors 10 are calculated from the structural model 20 by an orthogonal decomposition method using eigenvalues. A database is firstly constituted by grouping the results (in deformations and / or stresses) of the structural model 20 for numerous values ​​of the parameters (of the order of a few hundred). These values ​​are for example randomly drawn from the space of possible parameters. An orthogonal decomposition using eigenvalues ​​(also called principal component analysis) is then carried out to directly obtain all the vectors of the reduced basis. Finally, the positions of the sensors are the points maximizing each vector of the reduced basis.

[0080] In an exemplary embodiment, the reduced order model 12 and the measurement points of the sensors 10 are calculated automatically from the structural model 20 by a calculation module 22.

[0081] The first phase of the monitoring method comprises a third step of installing the sensors 10 at the determined measurement points of the marine structure 2.

[0082] The second phase of the monitoring method is carried out during operation of the marine structure 2, i.e. during use of the marine structure 2. It is carried out in situ and online.

[0083] The second phase of the monitoring method comprises in situ and online monitoring of the marine structure 2 by assimilation of the measurement signals provided by the sensors 10 by the reduced order model 12 configured to return a current state CS of the marine structure 2 from the sole knowledge of the measurement signals provided by the sensors 10, the current state CS of the marine structure 2 including estimation points distinct from the measurement points at which the sensors 10 are located.

[0084] The calculation of the current state CS of the marine structure 2 is carried out by the data-model hybridization model 14 from the measurements provided by the sensors 10 and the reduced order model 12.

[0085] Optionally, the second phase of the monitoring method advantageously comprises the calculation and transmission of monitoring indicators. This calculation and transmission are carried out by the processing and analysis module 16.

[0086] The calculation of the locations of the sensors 10 and the associated reduced order model 12 is implemented by computer.

[0087] The calculation module 22 is for example a software application comprising software code instructions recorded on a memory and executable by a processor.

[0088] The reduced order model 12 is presented in the form of a computer file suitable for being read and used by the data-model hybridization module 14. The MC measurements of the sensors 10 are provided in the form of files or signals suitable for being used by the data-model hybridization module 14.

[0089] The in situ and online monitoring of the marine structure 2 is implemented automatically by an electronic monitoring device 24.

[0090] The electronic monitoring device 24 comprises for example a memory 26 and a processor 28.

[0091] The data-model hybridization module 14 and, where appropriate, the analysis processing module 16 are, for example, each a software application comprising software code instructions recorded in the memory 26 and executable by the processor 28.

[0092] As a variant, at least one of the data-model hybridization module 14 and the processing and analysis module 16 is produced in the form of a programmable logic component or a specific integrated circuit (or ASIC for “Application Specific Integrated Circuit”).

[0093] The electronic monitoring device 24 is wholly or partly located on board in the marine structure 2.

[0094] In an exemplary embodiment, the electronic monitoring device 24 is entirely embedded in the marine structure 2.

[0095] Alternatively, the electronic monitoring device 24 is partly located at a distance from the marine structure 2, the monitoring being carried out remotely. The monitoring, however, remains carried out in situ and online because it is carried out from MC measurements taken by the sensors 10 on the marine structure 2 during its operation.

[0096] In an exemplary embodiment, the electronic monitoring device 24 is located in an installation separate from the maritime structure 2 and in communication with the maritime structure 2 via a telecommunications system. The MC measurements taken by the sensors 10 are then transmitted to the electronic monitoring device 24 via the telecommunications system.

[0097] The monitoring method makes it possible to provide in real time an estimate of the current state CS of the marine structure 2 to inform, for example, the crew operating the marine structure 2 or an operational maintenance team.

[0098] The monitoring method is based on the use of a reduced order 12 model which can be determined in a simple manner and easily implemented, with reduced computing power and short computing time.

[0099] The monitoring method can be deployed on any marine structure of interest requiring monitoring, including the structure of a surface or underwater vessel and the structure of a floating or stationary platform.

[0100] The monitoring method using sensors 10 arranged on the marine structure 2 and the reduced order model 12 specific to the assimilation does not require knowledge of the loading of the marine structure 2, which is very difficult to determine for the marine structures 2.

[0101] The monitoring method does not require the reduced order 12 model to be perfect, a reasonably representative reduced order 12 model being sufficient, the consideration of parametric and non-parametric uncertainties being integrated into the assimilation.

[0102] The monitoring method can be implemented with a reduced number of sensors 10.

[0103] The monitoring method makes it possible to obtain, in real time and at any point, the state of deformation and / or the state of stress of the marine structure 2, including in configurations not initially foreseen in the reduced order 12 model, by automatic correction of errors in the reduced order 12 model due to assimilation.

[0104] The monitoring method makes it possible to interpret any deviation from normal behavior and is not limited to detecting fatigue damage to the marine structure 2.

Claims

Claims

1. Method for monitoring a marine structure (2) using instrumentation (8) comprising sensors (10) positioned on the marine structure (2) at determined measurement points, the monitoring method comprising in situ and online monitoring of the marine structure (2) implemented by an electronic monitoring device (24) by assimilation of the measurements provided by the sensors (10) using a reduced order model (12) configured for calculating a current state (CS) of the marine structure (2) from the measurements provided by the sensors (10), the current state (CS) of the marine structure (2) including estimation points distinct from the measurement points where the sensors (10) are located.

2. A monitoring method according to claim 1, wherein the assimilation is performed by an empirical interpolation method.

3. A monitoring method according to claim 1, wherein the assimilation is performed by a parameterized low background data formulation method.

4. A monitoring method according to claim 1, wherein the assimilation is performed by a hybrid twin method.

5. A monitoring method according to any preceding claim, wherein each sensor (10) is a strain sensor or a strain sensor.

6. A monitoring method according to any preceding claim, wherein the current state (CS) calculated using the reduced order model (12) comprises the deformation and / or stress at said estimation points of the marine structure (2).

7. Monitoring method according to any one of the preceding claims, comprising the calculation, by the electronic monitoring device (24), of state indicators from the current state (CS) of the marine structure (2), and the transmission of the state indicators to a monitoring team.

8. A monitoring method according to claim 5, wherein the monitoring indicators comprise an estimate of the residual lifespan of the marine structure (2), an alert in the event of over-stressing of the marine structure (2), an alert in the event of abnormal behavior of the marine structure (2), a location of possible damage to the marine structure (2), the location of possible damage being possibly actually associated with a level of criticality.

9. A monitoring method according to any one of the preceding claims, wherein the marine structure (2) belongs to a surface vessel, a submarine vessel or a marine platform.

10. Monitoring method according to any one of the preceding claims, comprising: - obtaining a higher order structural model (20) representative of the marine structure (2), the structural model (20) being capable of receiving as input parameters and a loading of the marine structure (2) and of providing as output the deformations and / or the stresses of the marine structure (2) calculated as a function of the parameters and the loading received as input; and - calculating the reduced order model (12) and the locations (SP) for the sensors (10) from the structural model (20), of a set of constraints (CF).

11. The monitoring method of claim 10, wherein the reduced order model is calculated by an empirical interpolation method.

12. A monitoring method according to claim 10, wherein the reduced order model is calculated by an orthogonal eigenvalue decomposition method.

13. System for monitoring a marine structure (2), comprising a set of sensors (10) positioned on the marine structure (2) at measurement points and an electronic monitoring device (24) configured for implementing a method for monitoring the marine structure (2) according to any one of claims 1 to 9.

14. Computer program product recordable on a memory and executable by a processor, the computer program product being configured for the implementation of a method for monitoring a marine structure (2) according to any one of claims 1 to 9.

Images