EP2849994B1 - Method for predicting at least one movement of a ship under the effect of the waves - Google Patents

Description

The present invention relates to a method for predicting at least one movement of a ship lying on a body of water, under the effect of the swell of this body of water.

• un mouvement de translation du navire le long d'un axe longitudinal (également appelé cavalement),
• un mouvement de translation du navire le long d'un axe transversal (également appelé embardée),
• un mouvement de translation du navire le long d'un axe vertical (également appelé pilonnement),
• un mouvement de rotation du navire autour de l'axe longitudinal (également appelé roulis),
• un mouvement de rotation du navire autour de l'axe transversal (également appelé tangage), ou
• un mouvement de rotation du navire autour de l'axe vertical (également appelé lacet).

In the present description, the term "a movement of the ship" a translational movement along an axis, or rotation about an axis. In particular, the movement considered will generally be chosen from:

• a translational movement of the ship along a longitudinal axis (also called a caval),
• a translational movement of the ship along a transverse axis (also called yaw),
• a translational movement of the ship along a vertical axis (also called heave),
• a rotational movement of the ship about the longitudinal axis (also called roll),
• a rotational movement of the ship around the transverse axis (also called pitching), or
• a rotational movement of the ship around the vertical axis (also called yaw).

Certain operations carried out by the ship or from the ship, such as deployment or recovery operations of a drone, require a good stability of the ship. However, the swell causes the ship according to at least one of the aforementioned movements.

In order to be able to safely carry out an operation requiring a stability of the ship, it is necessary to predict the movements of the ship under the effect of the swell, to anticipate suitable moments for the realization of the operation and / or to compensate the movements induced by the swell.

For this purpose, it is already known in the state of the art, in particular according to DE 10 2007 011711 , US 2005/278094 or WO 03/075715 , various methods for predicting at least one movement of a ship under the effect of the swell. However, such methods of the state of the art generally do not allow to anticipate sufficiently in advance and sufficiently accurately the movements of the ship under the effect of the swell, or are very complex to implement.

The object of the invention is in particular to remedy these drawbacks, by providing a relatively simple prediction method, and allowing a sufficiently precise prediction and sufficiently in advance of the movements of the ship under the effect of the swell.

• une étape d'estimation d'une direction de propagation de la houle, et une étape d'estimation d'une vitesse de propagation de cette houle dans ladite direction de propagation,
• une étape de mesure de l'évolution d'une grandeur caractéristique de la houle, en au moins un point de mesure situé en amont du navire dans la direction de propagation, en mesurant périodiquement ladite grandeur,
• une étape de détection d'une accalmie de la houle au point de mesure, réalisée à l'aide de la mesure de l'évolution de la grandeur caractéristique, cette étape de détection comportant une mesure d'une durée d'une accalmie détectée, et

lorsqu'une accalmie de la houle est détectée au point de mesure :

• une étape de calcul d'un intervalle de temps entre la détection de l'accalmie de la houle au point de mesure détectée et un moment où cette accalmie se répercute sur le mouvement du navire, ce calcul étant notamment effectué en fonction de la vitesse de propagation estimée de la houle.

For this purpose, the invention particularly relates to a method for predicting at least one movement of a ship under the effect of the swell of a body of water, characterized in that it comprises:

• a step of estimating a wave propagation direction, and a step of estimating a propagation speed of this wave in said direction of propagation,
• a step of measuring the evolution of a characteristic magnitude of the swell, in at least one measuring point situated upstream of the ship in the direction of propagation, by periodically measuring said magnitude,
• a step of detecting a lull of the swell at the measurement point, carried out using the measurement of the evolution of the characteristic quantity, this detection step comprising a measurement of a duration of a detected lull, and

when a calm of the swell is detected at the point of measurement:

• a step of calculating a time interval between the detection of the calm of the swell at the measurement point detected and a moment when this lull has repercussions on the movement of the ship, this calculation being in particular carried out according to the speed of the estimated spread of the swell.

The invention proposes to measure the swell upstream of the ship, to detect the lulls of the swell upstream and to estimate the propagation of these lulls downstream to the ship, in order to predict the times when the ship is at the right of a calm of the swell.

This principle of the invention is based in particular on the fact that for sufficiently long lull periods, it is possible to neglect the deformation of the wave envelope between an upstream measurement point and the position of the downstream vessel. Thus, it is possible to consider only a single rate of lull propagation, rather than a different speed for each component of the wave spectrum.

Such a method is particularly simple to implement because it seeks simply to provide a lull for the movement of the ship, not to predict the precise behavior of this movement.

Indeed, it appears that for the carrying out of certain operations requiring a stability of the ship, it is sufficient to know a moment when the movement of the ship is weak (lull), without it being necessary to know the precise behavior of this ship . Thus, the method according to the invention has sufficient accuracy.

• Le procédé comporte, suite à l'étape de mesure, une étape de filtrage de l'évolution de la grandeur caractéristique mesurée, au moyen d'un filtre discret dont les entrées sont les grandeurs caractéristiques mesurées périodiquement, et les sorties formant un signal de sortie représentent l'effet de cette évolution de la grandeur caractéristique sur le mouvement considéré d'un navire fictif qui serait identique au navire et situé au point de mesure, et une étape de calcul d'une enveloppe du signal de sortie du filtre.
• L'étape de détection d'une accalmie de la houle au point de mesure comprend : une comparaison de l'enveloppe avec un seuil d'amplitude prédéterminé, et la mesure de la durée d'accalmie, réalisée en mesurant la durée pendant laquelle l'enveloppe est inférieure audit seuil d'amplitude, une accalmie étant considérée comme détectée lorsque ladite durée d'accalmie est supérieure à un premier seuil de durée prédéterminé.
• L'étape de calcul d'une enveloppe comporte l'application d'une transformée de Hilbert au signal de sortie de filtre.
• La transformée de Hilbert est effectuée sur une fenêtre glissante appliquée au signal de sortie de filtre, la fenêtre glissante étant choisie pour coïncider entre deux passages par 0.
• Le procédé comporte, suite à l'étape de calcul d'une enveloppe et préalablement à l'étape de détection d'une accalmie, une étape de décomposition de l'enveloppe en ondelettes.
• Les ondelettes sont des ondelettes de Meyer.
• L'étape de filtrage est réalisée au moyen d'un filtre discret linéaire et causal, et présentant la forme suivante : $$s\left(t_i\right)=C\cdot X\left(t_i\right)+D\cdot h\left(t_i\right)$$
  
  avec :
  • h(t_i ) la grandeur caractéristique de la houle à un instant de mesure t_i ,
  • s(t_i ) la valeur du signal de sortie de filtre à l'instant de mesure t_i ,
  • X(t_i ) une fonction matrice causale de la forme X(t _i+1) = A·X(t_i )+B·h(t_i ), avec X(t ₀)=0, et
  • A, B, C et D des matrices constantes.
• Le procédé comporte, suite à l'étape de calcul de l'intervalle de temps, une étape d'estimation d'une probabilité que le mouvement du navire sous l'effet de la houle, au moment où l'accalmie détectée se répercute sur le mouvement du navire, soit inférieur à un seuil de mouvement prédéterminé pendant une durée supérieure à un second seuil de durée prédéterminé, cette estimation étant notamment effectué en fonction de la durée de l'accalmie détectée.
• La grandeur caractéristique de la houle est choisi parmi une élévation de la surface de l'étendue d'eau au point de mesure, une vitesse d'élévation de la surface de l'étendue d'eau au point de mesure, ou une pression de l'eau au point de mesure.

The method according to the invention may further comprise one or more of the following characteristics, taken alone or in any technically possible combination.

• The method comprises, after the measurement step, a step of filtering the evolution of the measured characteristic quantity, by means of a discrete filter whose inputs are the characteristic quantities measured periodically, and the outputs forming a signal of output represent the effect of this change in the characteristic magnitude on the considered motion of a fictitious ship that would be identical to the ship and located at the measurement point, and a step of calculating an envelope of the output signal of the filter.
• The step of detecting a lull of the swell at the measurement point comprises: a comparison of the envelope with a predetermined amplitude threshold, and the measurement of the duration of lull, carried out by measuring the duration during which the envelope is below said amplitude threshold, a lull being considered as detected when said lull period is greater than a first threshold of predetermined duration.
• The step of calculating an envelope comprises applying a Hilbert transform to the filter output signal.
• The Hilbert transform is performed on a sliding window applied to the filter output signal, the sliding window being selected to coincide between two passes by 0.
• The method comprises, following the step of calculating an envelope and prior to the step of detecting a lull, a step of decomposing the wavelet envelope.
• The wavelets are Meyer wavelets.
• The filtering step is performed by means of a linear discrete and causal filter, and having the following form: $$s\left(t_i\right)=\mathrm{VS}\cdot X\left(t_i\right)+D\cdot h\left(t_i\right)$$
  
  with:
  • h ( t _i ) the characteristic quantity of the swell at a measurement instant t _i ,
  • s ( t _i ) the value of the filter output signal at the measurement instant t _i ,
  • X ( t _i ) a causal matrix function of the form X ( t _{i + 1} ) = A · X ( t _i ) + B · h ( t _i ), with X ( t ₀ ) = 0, and
  • A, B, C and D constant matrices.
• The method comprises, following the step of calculating the time interval, a step of estimating a probability that the movement of the ship under the effect of the swell, at the moment when the lull detected has an effect on the the movement of the ship is less than a predetermined movement threshold for a duration greater than a second threshold of predetermined duration, this estimate being made in particular according to the duration of the lull detected.
• The characteristic magnitude of the swell is selected from an elevation of the surface of the body of water at the measuring point, a rate of elevation of the surface of the body of water at the point of measurement, or a pressure of water at the measuring point.

• la figure 1 un navire se trouvant sur une étendue d'eau ;
• la figure 2 représente les étapes du procédé selon l'invention, pour la prévision d'au moins un mouvement du navire de la figure 1.

The invention will be better understood on reading the description which follows, given solely by way of example and with reference to the appended figures, among which:

• the figure 1 a ship lying on a body of water;
• the figure 2 represents the steps of the method according to the invention, for the purpose of predicting at least one movement of the ship of the figure 1 .

We have shown on the figure 1 a ship N being on a body of water, intended to carry out at least one operation requiring a stability of the ship, such as a deployment operation or recovery of a drone.

In order to provide such stability of the ship N, it is necessary to provide at least one movement of the ship N under the effect of the swell of the body of water, and thus provide a time when this movement is weak.

The movement considered is chosen from the caval, the lurching, the heave, the roll, the pitch or the yaw of the ship N.

For this purpose, it is shown on the figure 2 the steps of a method for predicting at least one movement of the ship N under the effect of the swell of the body of water, according to an exemplary embodiment of the invention.

The method according to the invention comprises a preliminary step 10 for estimating a direction D of propagation of the swell, as well as a step 20 for estimating a propagation velocity of this swell in said direction of propagation D.

These estimation steps 10, 20 are performed using means for estimating the direction and velocity of propagation of the swell. Such means are known per se, and will not be described in detail. For example, these estimation means comprise a RADAR type monitoring system known per se, carried by the ship N, and / or adapted buoys arranged on the body of water to carry out measurements, and suitable for communicating these. measurements to the ship N.

These estimation steps 10, 20 can be performed again at any time of the forecasting method, so as necessary, to provide an update of the direction D and the speed of propagation of the swell.

Once the wave propagation direction D of known waves, the method comprises a step 30 of measuring the evolution of a magnitude characteristic of the swell, in at least one measurement point P upstream of the ship N in the direction of propagation D, as shown on the figure 1 . The evolution of the quantity is measured by periodically measuring said quantity.

The measured quantity may be any characteristic magnitude of the swell, making it possible to obtain the instantaneous wave energy, for example the elevation of the free surface, the elevation speed of this free surface, or the pressure at a height. determined. Note that the measurement can be performed at a point P, or on a defined space, for example on a measurement grid.

These measurements can be made using means known per se, such as a RADAR type system, LIDAR type, or other, carried by the ship, or by adapted buoys arranged on the body of water for carry out the measurements and to communicate these measurements to the ship N.

où

• t ₀ est l'instant de la première mesure effectuée,
• dt est la période à laquelle sont effectuées les mesures, dite période d'échantillonage, et
• i est le rang de la mesure considérée.

In what follows, we will denote h ( t _i ) a measure of the magnitude carried out at a given moment $$t_i=t_0+i\cdot \mathit{dt},$$

or

• t ₀ is the moment of the first measurement made,
• dt is the period during which measurements are taken, called the sampling period, and
• i is the rank of the measure considered.

The set of periodic measurements of the quantity forms a discrete sequence, representing the evolution of this measured characteristic quantity.

The method then comprises a step 40 of filtering the evolution of the measured characteristic quantity, by means of a discrete filter whose inputs are the measurements h ( t _i ) of the periodically measured characteristic quantity, and the outputs represent the effect of the evolution of this characteristic quantity on the movement considered.

It will be noted that, if one wishes to study several movements of the ship, it is necessary to provide as many filters as movements considered, in order to study the effect of the swell on each of these movements.

It should also be noted that the measurements being made upstream of the ship, the outputs of the filter correspond to the fictitious movement, under the effect of the swell, of a fictitious ship (designated by the reference N 'on the figure 1 ), with the same characteristics that the ship N, which would be located at the point of measurement P. In the remainder of the present description, the signal formed by outputs of this filter will be called "upstream dummy movement".

In what follows, the output of the filter at a time t _i will be noted s ( t _i ).

où

• X est une fonction vectorielle causale telle que : X(t _i+1) = A·X(t_i )+B·h(t_i ), avec X(t ₀) = 0, et
• A, B, C et D sont des matrices constantes.

According to the embodiment described, the chosen filter is a linear discrete and causal filter, having the following form: $$s\left(t_i\right)=\mathrm{VS}\cdot X\left(t_i\right)+D\cdot h\left(t_i\right),$$

or

• X is a causal vector function such that: X ( t _{i + 1} ) = A · X ( t _i ) + B · h ( t _i ), with X ( t ₀ ) = 0, and
• A, B, C and D are constant matrices.

The values of the constant matrices A, B, C and D are determined experimentally, and are chosen to minimize the difference between the real movements of the ship in response to the swell, and the fictitious movements reconstituted by this filter. In particular, these values are a function of the characteristics of the ship, the speed of that ship and the impact of the swell in relation to the heading of that ship, as well as the vessel movement considered.

When the movement considered is the roll of the ship, the filter is for example of order 4, namely a first filter of order 2 for approximating the natural mechanical resonance of the ship, and a second filter of order 2 in cascade allowing to approximate the excitation at the moment of roll generated by the swell.

In order to study the upstream fictitious motion signal, the method comprises a step 50 of calculating an envelope of this upstream fictitious motion signal. For this purpose, applying a Hilbert transform H (s (t)) to the output signal of filter s (t) to obtain the imaginary part of an _analytical analytic signal S (t).

Et $$S_{\mathit{analytique}}\left(t\right)=s\left(t\right)+i\cdot H\left(s\left(t\right)\right)$$

So : $$\mathrm{H}\left(s\left(t\right)\right)=\frac{1}{\pi }{\displaystyle {\int }_{-\infty }^{+\infty }}\frac{s\left(\tau \right)}{t-\tau }\mathit{d\tau },$$

And $$S_{\mathit{analytic}}\left(t\right)=s\left(t\right)+i\cdot H\left(s\left(t\right)\right)$$

The envelope of the signal s (t), denoted S _env ( t ), is the norm of the analytical signal. $$S_{\mathit{ca.}}\left(t\right)=\left|S_{\mathit{analytic}}\left(t\right)\right|$$

1. a) on calcule la transformée de Fourier rapide du signal S(f)= FFT(s(t))
2. b) on calcule un signal S'(f) défini de la manière suivante :
   • pour les fréquences f positives, S'(f) = 2 × S(f)
   • pour les fréquences f négatives, S'(f)=0,
   • pour la fréquence nulle et la fréquence de Shannon, S'(f)=S(f).
3. c) on calcule la transformée inverse du signal S'(f), et on obtient ainsi l'enveloppe S_env (t)=IFFT(S'(f)).

In the case of a discrete signal s (t), the envelope is calculated by the following algorithm:

1. a) the fast Fourier transform of the signal S ( f ) = FFT ( s ( t )) is calculated
2. b) calculating a signal S ' ( f ) defined as follows:
   • for the positive frequencies f, S ' ( f ) = 2 × S ( f )
   • for the negative frequencies f, S ' ( f ) = 0,
   • for the zero frequency and the Shannon frequency, S ' ( f ) = S ( f ).
3. c) calculating the inverse transform of the signal S ' ( f ), and thus obtaining the envelope S _env ( t ) = IFFT ( S' ( f )).

Preferably, the Hilbert transform is performed on a sliding window applied to the filter output signal. Advantageously, the sliding window is chosen to coincide between two zero crossings, that is to say that s (t) = 0 at the input and at the output of the window. The signal is then extended by a mirror operation which ensures a continuity of the periodic function and its derivative, and thus attenuates the windowing effects. This mirror operation, known per se, consists in considering that the signal upstream or downstream of the window is symmetrical to the signal inside the window, with respect to the point of the signal at the input, respectively at the exit, from the window.

Indeed, by applying a simple rectangular window, without any upstream processing, artifacts (also called edge effects) appear on the edges of the signal. On the other hand, if one performs the mirror operation before applying the Hilbert transform in the window, the discontinuities disappear.

Thanks to the envelope obtained, it will be possible to detect a lull of the swell, compared to the movement considered, that is to say a lull of the swell which causes only a movement considered sufficiently weak. For this purpose, the method comprises a step 60 of decomposition of the wavelet envelope, which makes it possible to isolate the lowest frequency components of this envelope.

The number of components to be considered can be predetermined, or established on a criterion of fractions of the energy. For example, we can use Meyer wavelets.

The method then comprises a step 70 of detecting a lull of the swell at the measurement point P, this step being performed from the obtained wavelets.

During this detection step 70, the envelope is compared with a predetermined amplitude threshold.

This detection step 70 also provides a measurement of a lull period, that is to say a duration during which this envelope is less than said predetermined amplitude threshold.

It is then considered that a lull is detected when the measured lull time is greater than a first threshold of predetermined duration.

When such a lull is detected, it can be considered that it propagates in the direction D of propagation of the swell, the speed of propagation of this swell, so in the direction of the ship N.

The method therefore then comprises a step 80 for calculating a time interval between the detection of the calm of the swell at the measurement point P and a moment when this lull has repercussions on the movement of the ship N. This calculation is notably performed as a function of the wave propagation velocity previously estimated during the estimation step 20.

It should be noted that the calculation of the time interval also depends on the distance from the point P with respect to the ship N. Thus, if it is desired to have a time interval large enough to prepare the operation, it will be possible to choose a point P further away.

When an operation of the ship requires its stability in several movements, it is considered that this operation can be performed when a lull will be detected simultaneously for each of these movements.

It will be noted that it may occur that a lull does not propagate from the measurement point P to the ship N, especially when this measuring point P is particularly far from the ship N. Thus, the method preferably comprises in step 80 of calculating the time interval, a step 90 of calculating a probability so that the detected lull has an effect on the movement of the ship, that is to say so that this movement of the vessel under the effect of the swell is less than a predetermined movement threshold for a duration greater than a second threshold of predetermined duration.

This second threshold of predetermined duration corresponds to the minimum time necessary to perform the operation.

This calculation is made in particular according to the duration of the lull detected. This estimation of probability can be performed by calculation, using the theory of detection, known per se, detection probability formulas, and false alarms. Alternatively, the probability estimate can be made by learning, this learning can for example be performed by counting, for a determined number of lulls detected, how much is propagated to the ship, in order to deduce a percentage.

Examples of probabilities obtained during tests of the process according to the invention are shown in the table below.

In particular, a first threshold of duration (duration of a lull at point P) of 50 seconds was considered, and a second threshold of duration (duration during which the movement of the ship is less than the predetermined threshold of movement) of 40 seconds. .

• la première colonne précise la distance du point P au navire N, en mètres
• chaque double colonne concerne un exemple de mouvement particulier, et comporte :
  • o une colonne indiquant l'intervalle de durée mesurée entre l'accalmie au point P et l'accalmie au navire N, en secondes
  • o une colonne indiquant la probabilité d'une accalmie d'au moins 40 seconde au navire lorsqu'une accalmie d'au moins 50 secondes a été détectée au point P, en %.

So, in the table below:

• the first column specifies the distance from point P to ship N, in meters
• each double column concerns an example of a particular movement, and includes:
  • o a column indicating the measured time interval between the lull at point P and the lull at ship N, in seconds
  • o a column indicating the probability of a lull of at least 40 seconds to the vessel when a lull of at least 50 seconds has been detected at point P, in%.

The movements considered are heave, rolling and pitching. Indeed, a lull of these three movements is generally necessary for a deployment operation or recovery of a drone. heave Roll Pitch Distance Probability Time Probability Time Probability Time 480 m 90% 18 s 98% 13s 90% 31 s 720 m 50% 47 s 90% 40s 90% 66 s 960 m 55% 76 s 90% 67 s 70% 101 s

It is clear that the further away the point P is, the lower the probability of a lull to the ship, but the longer the time to prepare the operation. The distance of the ship to the point P will therefore generally be chosen according to the best compromise between the need for a significant time interval to prepare the mission and the wish for a sufficient probability of calm.

Note that the invention is not limited to the embodiment described above, and could have various variants without departing from the scope of the claims.

Claims (10)

1. A prediction method for predicting at least one movement of a ship (N) on an area of water under the effect of a swell on this area of water, characterised in that it includes:

   - a step (10) of estimating a direction (D) of propagation of the swell, and a step (20) of estimating a propagation speed of the swell in the direction (D) of propagation,

   - a step (30) of measuring the development of a characteristic value of the swell at at least one measuring point (P) upstream of the ship in the direction of propagation (D) by periodically measuring the value,

   - a step (70) of detecting a lull in the swell at the measuring point (P) using a measurement of the development of the characteristic value, including a measurement of a duration of a lull detected, and

   if a lull in the swell is detected at the measuring point (P):

   - a step (80) of calculating a time interval between the detection of the lull in the swell at the detected measurement point (P) and a moment in which the lull affects the movement of the ship (N), carried out, in particular, depending on the estimated speed of propagation of the swell.
2. Prediction method according to claim 1, including:

   - following the measuring step (30), a step (40) of filtering the development of the characteristic value measured by means of a discrete filter, the inputs of which are the characteristic value periodically measured and the outputs of which are an output signal representing the effect of the development of the characteristic value on the movement in question of a notional ship (N') identical to the ship (N) and positioned at the measurement point (P),

   - a step (50) of calculating an envelope of the output signal of the filter.
3. Prediction method according to claim 2, in which the step (70) of detecting a lull in the swell at the measurement point (P) comprises:

   - a comparison of the envelope with a predetermined amplitude threshold,

   - the measurement of the duration of the lull by measuring the duration in which the envelope is less than the amplitude threshold,

   whereby a lull is considered to have been detected if the measured duration of the lull is greater than a first predetermined time threshold.
4. Prediction method according to claim 2 or 3, in which the step (50) of calculating an envelope includes the application of a Hilbert transformation to the output signal of the filter.
5. Prediction method according to claim 4, in which the Hilbert transformation is carried out on a sliding window applied to the output signal of the filter, whereby the sliding window is chosen to coincide between two 0 passes.
6. Prediction method according to any of claims 2 - 5, including, following the step (50) of calculating an envelope and before the step (70) of detecting a lull, a step (60) of breaking down the envelope into wavelets.
7. Prediction method according to claim 6, in which the wavelets are Meyer wavelets.
8. Prediction method according to any of claims 2 - 7, in which the step (40) of filtering is carried out by means of a discrete linear, causal filter having the following form: $$s\left(t_i\right)=C\cdot X\left(t_i\right)+D\cdot h\left(t_i\right)$$

   

   where:

   h(t_i ) is the characteristic value of the swell at a measurement time t_i ,

   s(t_i ) is the value of the output signal of the filter at the time of measurement t_i ,

   X(t_i ) is a causal matrix function having the form X(t _i+1) = A · X(t_i ) + B · h(t_i ), where X(t ₀)=0, and

   A, B, C, and D are constant matrices.
9. Prediction method according to any of the foregoing claims, including, following the step (80) of calculating the time interval, a step (90) of estimating a probability that the movement of the ship (N) under the effects of the swell, when the lull detected affects the movement of the ship (N), is less than a predetermined movement threshold for a duration greater than a second predetermined time threshold, whereby the estimation is carried out, in particular, depending on the duration of the lull detected.
10. Prediction method according to any of the foregoing claims, in which the characteristic value of the swell is selected from an elevation of the surface of the area of water at the measurement point, a speed of elevation of the surface of the area of water at the measurement point, or a water pressure at the measurement point.

Images