Method and device for managing a magnetic signature of a ship, corresponding computer program product and corresponding storage medium

Abstract

The invention relates to a method for managing a magnetic signature of a ship, carried out by a computing machine and comprising a step (20) of magnetically modelling the ship, comprising: obtaining (21) magnetic field measurements each associated with information comprising a geographical position, a magnetic heading of the ship and a distance from the ship; and determining (22) a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information. The first inverse problem is solved with a first solution method based on sparsity regularization using an a priori that favours sparsity. The magnetic moment model of the ship is used in a step (23) of predicting the magnetic signature of the ship and / or in a step (24) of compensating for the magnetic signature of the ship.

Description

[0001]DESCRIPTION TITLE: Method and device for managing a magnetic signature of a ship, corresponding computer program product and storage medium 1. TECHNICAL FIELD The field of the invention is that of the magnetism of ships (naval vessels). The invention relates more particularly to a method and device for managing a magnetic signature of a ship. The invention thus has numerous applications, in particular, but not exclusively, the prediction of a ship's signature and the compensation of a ship's signature. Generally speaking, the magnetic signature of a surface ship corresponds to the total field on a reference line or plane located under the keel at a distance equal to the width of the ship. It depends on the position of the ship on the globe and its orientation. It represents the deformation of the Earth's magnetic field by the magnetization of the ship.Knowing and reducing the magnetic signature of a ship (especially a military one) is crucial to ensure its safety when approaching the coast (potentially mined area) and, in the particular case of a submarine, it is its discretion that is directly impacted (detection by an aircraft / drone). It is therefore essential to have a means of measuring this magnetic signature. This measurement is however not sufficient and a modeling test is essential to be able to predict the magnetic signature at a different location (geographical position) on the globe and / or at a different distance from the ship and / or at a different attitude of the ship, the magnetic signature depending among other things on the local Earth magnetic field and the orientation of the ship in it. The attitude of the ship is defined by the roll, pitch and magnetic heading, which correspond to three rotations around the three conventional orthogonal axes defined for the ship.More precisely, the ship's coordinates are related to a direct orthogonal trihedron whose x-axis is oriented towards the bow, the y-axis towards starboard and the z-axis downwards. In such a system, roll corresponds to rotations around the x-axis (positive when the ship is listing to starboard), pitch corresponds to rotations around the y-axis (positive when the bow is pointing upwards) and heading variations (also called "yaws") correspond to rotations around the z-axis (positive clockwise). The magnetic heading (also called geomagnetic heading) of a ship is the angle expressed in degrees (from 0 to 360°), in a clockwise direction, between the direction of magnetic north and the ship's heading line (i.e. the direction in which the ship's bow is pointing, according to the longitudinal axis of the ship).A ship's magnetic heading can be obtained either directly (e.g., with a magnetometer) or indirectly (e.g., by calculation from the compass heading or the geographic heading). A ship's compass heading is the heading indicated by the compass, i.e., the angle between compass north and the ship's heading line. The angular difference between compass north and magnetic north is called deviation. A ship's geographic heading (also called true heading) is the angle between geographic north (also called true north) and the ship's heading line. The angular difference between geographic north and magnetic north is called declination. Beyond predicting the ship's magnetic signature, knowing it makes it possible to determine what actions should be taken to reduce it. This can be done, in particular, by an immunization system (also called a "degaussing system") on board the ship and consisting of current loops.To ensure the reduction of the magnetic signature, it is necessary to adjust the immunization system (usually this is done on a measuring station) and, again, by modeling, correctly distribute the loop currents (also called "degaussing currents") according to the predicted signature (function of the location limits (Earth field), distance and attitude). 2. TECHNOLOGICAL BACKGROUND The techniques currently used make it possible to determine a magnetic model of the ship in magnetic moments, by solving an inverse problem (i.e. a so-called "inverse" mathematical problem) aimed at determining the ship's magnetic moments from magnetic field measurements and associated information (geographical position of the sensor carrying out the measurement, magnetic heading of the ship and a distance from the ship).Such determination of a magnetic model of the ship by solving an inverse problem (also called "inverse modeling" or "modeling according to the inverse model") is described for example in patent application WO2017064432A1. But the mathematical methods used (in particular the least squares method) do not lead to a unique solution because the problem is ill-posed (small variations in the measured fields can drastically change the calculated solution). Indeed, the philosophy of modeling the ship and adjusting the degaussing system and the logic of interpretation, used for decades by experts in the field of ship magnetism, have never been explicitly translated mathematically into the field of magnetic moments for solving the problem of modeling and adjusting the degaussing system.Indeed, the ship is commonly modeled by a plane of magnetic moments (dipoles; elementary sources of magnetism) and the mathematical methods used attempt to find the value of each of these moments. More precisely, in the state of the art, the usual methods seek to find the value and direction of each of the equally distributed sources by minimizing a criterion between the real magnetic field measurement and the reconstructed measurement, considering that all potential sources have an existence (i.e. a non-zero value). The solution found is thus mathematically correct but has very little chance of having a physical meaning (depends a lot on the type of regularization used to solve the inverse problem). However, there is only one solution that satisfies physics and this problem requires going back to the domain of magnetic fields to predict the signature and adjust the degaussing system.There is therefore a risk of misinterpretation of the magnetic field (in particular, there is a risk of compensating a magnetic moment along one axis by degaussing loops along the other axes) and this reserves the exploitation of the prediction and adjustment of the degaussing system to a handful of experts around the world. Without an expert, and this is particularly the case for operating a portable measuring station on board the ship, it is no longer possible to correctly exploit the measurements. In summary, current models suffer from giving a mathematically correct solution but this solution is not unique, it has no reason to be close to physical reality and this constrains the ability to predict the magnetic signature of the ship and the risk of detection to an area close to the initial measurement and, for the adjustment of the degaussing system, does not allow simple automatic adjustment or adjustment outside this same area.The use of different mathematical methods (Bayesian methods, in particular) has begun for a few years, but for the moment, these methods coming from statistics have been exploited just for the additional information they provide (standard deviation, error estimation, etc.). The authors continue to look for magnetic moments equally distributed in the modeling plane and to return to the domain of magnetic fields for signature prediction and adjustment of the degaussing system. In order to illustrate the fact that modeling, signature prediction and adjustment of degaussing currents today inexorably pass through the domain of magnetic fields, we now present, in relation to Figure 1, a method for managing a magnetic signature of a ship according to a particular embodiment of the prior art.In a step 11, for a given magnetic heading, the device (executing the method) obtains magnetic field measurements each associated with information comprising a geographical position, the given magnetic heading of the ship and a distance from the ship. In a step 12, for the aforementioned given magnetic heading (see step 11), the device determines a magnetic model of the ship in magnetic moments, by solving an inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information. In a step 13, for the aforementioned given magnetic heading (see steps 11 and 12), the device determines a magnetic signature of the ship, as a function of the magnetic model of the ship. In other words, the device returns to the domain of magnetic fields.The magnetic signature determined in step 13 is valid only for the set of information associated with the magnetic field measurements (same geographical position, same magnetic heading of the ship and same distance from the ship). As indicated by the frame referenced 10, steps 11, 12 and 13 are iterated for each of a plurality of N magnetic headings (generally the following four (N = 4) particular magnetic headings are considered: north magnetic heading (when the magnetic heading is equal to 0°), south magnetic heading (when the magnetic heading is equal to 180°), east magnetic heading (when the magnetic heading is equal to 90°) and west magnetic heading (when the magnetic heading is equal to 270°)). A magnetic signature of the ship is thus obtained for each of this plurality of magnetic headings. In other words, a “signature at different magnetic headings” is obtained (if N is greater than 1).In a step 14, the device makes a prediction of the magnetic signature of the ship at a geographical position and / or a magnetic heading of the ship and / or a distance from the ship other than that or those of the magnetic field measurements, as a function of the magnetic signature (magnetic field domain) for different magnetic headings (result of the N iterations of step 13 for the N different magnetic headings) and using charts and estimators constructed by experimentation. In a step 15, the device makes a compensation of the magnetic signature of the ship using the degaussing system (immunization system comprising current loops installed on board the ship).The compensation step 15 itself comprises the following steps: - determination in step 16 of the magnetic effect, in the magnetic field domain, of the current loops; - for each of the current loops, determination in step 17 of a compensation current as a function of the magnetic signature of the ship (that determined at different magnetic headings in step 13, or that predicted in step 14; the second case is symbolized by the arrow referenced 19) and of the magnetic effect (in magnetic fields) of the current loops. In other words, step 17 comprises analysis and calculation operations carried out in the magnetic field domain; and - application in step 18, on the current loops, of the determined compensation currents.For more details on compensating for a ship's magnetic signature using a degaussing system, please refer, for example, to patent application WO2017064432A1 already mentioned above, as well as patents EP0597014B1 and US6965505B1. 3.SUMMARY In one embodiment of the invention, a method for managing a magnetic signature of a ship is proposed, executed by a computing machine and comprising a step of magnetic modeling of the ship itself comprising: - obtaining magnetic field measurements each associated with information comprising a geographical position, a magnetic heading of the ship and a distance from the ship; and - determining a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information; the resolution of the first inverse problem being carried out with a first resolution method based on a sparse regularization using an a priori favoring sparseness. Thus, the proposed solution offers a completely new and inventive approach.Indeed, as indicated above, the modeling, the signature prediction and the adjustment of the degaussing currents always pass, in the known solutions of the prior art, through the domain of magnetic fields. However, the magnetic field is a consequence and not the magnetic source, the magnetic field being what can be measured, but it introduces complexities and biases. The object of the invention is precisely to be able to have access to a model of the ship in magnetic sources, that is to say in magnetic moments, usable to carry out operations directly in this domain of magnetic moments. The challenge is in fact that a magnetic moment has a given direction while the resulting magnetic field is along the three axes depending on the place where one stands to measure.The objective is to correctly know the direction of each of the magnetic sources to correctly predict the magnetic field created by these sources and thus correctly perform the signature prediction and / or the adjustment of the degaussing currents. The proposed solution therefore proposes to use a sparse regularization method to determine the magnetic moment model of the ship. Because it uses a priori favoring sparseness, this sparse regularization method makes it possible to constrain the resolution of the first inverse problem (that aiming to determine the magnetic moments of the ship from the magnetic field measurements and associated information) so that the magnetic model of the ship, obtained in magnetic moments, can be directly used in a device usable by any person on board the ship.In other words, the idea of ​​the invention is that the simplest solution is also the most physical: there is no point in trying to put magnetic sources everywhere in the model, equally distributed, it is better to look for a smaller number (preferably the minimum) of sources which explain the measurements and therefore allow oneself to cancel certain sources (this is also consistent with the design principle of a degaussing system, a few loops therefore magnetic moments sufficient to compensate a ship). This is an essential difference with the state of the art, since the use of a so-called parsimonious method aims precisely to model the ship by a limited number of magnetic sources (magnetic moments). According to a particular characteristic, the first based on a. sparse regularization on at least one of the norms ℓ ^ and ℓ ^ . The advantage of these standards ℓ ^ and ℓ^ is to allow finding a restricted number of magnetic moments (parsimony property). In this description, the norm ℓ ^ of the vector ^ (which is written ∥ ^ ∥ ^ ) counts the number of non-zero elements of the vector ^. Note that this is an abuse of language, and that strictly speaking the correct writing is "norm" ℓ ^ , because it is actually a pseudo-norm. The norm ℓ ^ of a vector ^ = (^ ^ , …, ^ ^ ) is written: ∥ ^ ∥ ^ ^= (∑ ^^^ | ^^ |) . More we recall that for ^ ≥ 1, the norm ℓ ^ of a vector ^ = (^ ^ , …, ^ ^ ) ^ ^ the first inverse problem is written in one of the following mathematical forms: ^^^^^^ ^^^^^^ ^^^^^^ ^ ^ ^ with ^ ^^^^^^ a vector of the ship's magnetic moments to be estimated, ^ ^^^^^^a vector of magnetic field measurements and ^ a matrix of position-dependent coefficients associated with the magnetic field measurements. This list of mathematical forms is not exhaustive. In a particular implementation, the first solution method based on sparse regularization is a Bayesian method. An advantage of the Bayesian method is that it provides a good compromise between convergence speed and estimation quality of the vector ^ ^^^^^^, and also provides the uncertainty on the estimated magnetic moments. However, the present invention is not limited to this type of method. According to certain embodiments, the method comprises a step of predicting the magnetic signature of the ship at a geographical position and / or a magnetic heading of the ship and / or a distance from the ship other than that, that or those of the magnetic field measurements, according to the magnetic moment model of the ship.Thus, the ship's magnetic moment model (obtained at the modeling stage) makes it possible to predict the ship's magnetic signature under conditions different from those of the magnetic field measurement (i.e. at a given magnetic heading and / or at a given distance and / or at a given location other than those of the measurement), and thus to give sailors on board the ship an estimate of their detection distance by an enemy device (approach distance, minimum navigation depth or depth for a ship and diving direction for a submarine). It should be noted that, unlike the state of the art, the proposed solution allows a prediction with a change in distance.Indeed, because, in the invention, the magnetic moment model of the ship results from the use of a sparse regularization method, it is sufficiently close to physical reality (unlike ship models obtained with state-of-the-art solutions) to have validity outside the initial measurement zone (within the validity limit of a degaussing system or, in other words, outside the so-called near-field zone).According to certain embodiments, the method comprises a step of compensating the magnetic signature of the ship using an immunization system comprising current loops installed on board the ship, the step of compensating the magnetic signature of the ship comprising: - determining a magnetic moment(s) model of each of the current loops; - for each of the current loops, determining a compensation current as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops; and - applying, to the current loops, the determined compensation currents. Thus, the magnetic moment model of the ship (obtained in the modeling step) makes it possible to perform compensation of the magnetic signature of the ship using an immunization system.As already mentioned above, in the invention, the magnetic moment model of the ship results from the use of a sparse regularization method, it is sufficiently close to physical reality (unlike ship models obtained with state-of-the-art solutions) to have validity outside the initial measurement area. Since the magnetic moments are very directly linked to the electric currents, and therefore to the loop effects, they offer the advantage of being able to adjust the current loops more easily and above all automatically, making a lambda operator autonomous. The compensation of the magnetic signature of the ship can therefore be carried out in the form of an automatic adjustment of the degaussing currents, with physical quantities in "amperes x meters²" which correspond directly to the characteristics of the current loops (magnetic compensation loops).Ultimately, the methodology provides the operator with currents (in amperes) to be applied to the existing compensation loops. According to a particular characteristic, the calculation of the compensation currents is carried out by solving a second inverse problem, aiming to determine the compensation currents according to the magnetic moment model of the ship and the magnetic moment models of the current loops, with a second resolution method based on a sparse regularization using an a priori favoring sparseness. Thus, it is proposed to use a sparse regularization method also to determine the compensation currents (degaussing currents).Because it uses a priori favoring sparsity, such a sparse regularization method makes it possible to constrain the resolution of the second inverse problem (the one aiming to determine the compensation currents as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops) so that the solution is physically relevant. According to a particular characteristic, the second resolution method based on sparse regularization relies on at least one of the ℓ norms. As already mentioned above, the advantage of these standards ℓ ^ is to facilitate calculations. According to certain embodiments, the second inverse problem is written in one of the following mathematical forms: ^ ^^^^^^ 1 ^^^^^^^ ^^^ ∥ ^ ∥ ^ under the constraint ^ = − ^ ^ ^^^ ^ ^^^^ 1 ^^^ ∥ ^ ^ ^^^^^^^ ^ ∥ ^ under the constraint ^ = − ^ ^ ^ ^^^ ^ ^ ^ ^ ^^^ with ^ a vector of compensation currents to be estimated, ^ ^^^^^^^ a matrix of coefficients depending on the magnetic models of the current loops and ^ ^^^a reference current (applied to each of the compensation loops when measuring the effects thereof). This list of mathematical forms is not exhaustive. In a particular implementation, the second resolution method based on sparse regularization uses a Markov Chain Monte Carlo type algorithm. The present invention is however not limited to this type of method. In another embodiment of the invention, a computer program product is provided which comprises program code instructions for implementing the aforementioned method (in any of its various embodiments), when said program is executed on a computer.In another embodiment of the invention, there is provided a non-transitory, computer-readable storage medium storing a computer program product comprising a set of computer-executable instructions for implementing the above-mentioned method (in any of its various embodiments).In another embodiment of the invention, a device for managing a magnetic signature of a ship is proposed, the device comprising a dedicated or reprogrammable computing machine, configured to perform a step of magnetic modeling of the ship itself comprising: - obtaining magnetic field measurements each associated with information comprising a geographical position, a magnetic heading of the ship and a distance from the ship; and - determining a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information; the resolution of the first inverse problem being carried out with a first resolution method based on a sparse regularization using an a priori favoring sparseness.In embodiments of the device for managing a magnetic signature of a ship, the calculation machine is also configured for the implementation of other step(s) (in particular the prediction step and / or the compensation step) and / or characteristic(s) of the aforementioned method, in any of its different embodiments. 4. LIST OF FIGURES Other characteristics and advantages of the invention will appear on reading the following description, given by way of illustrative and non-limiting example, in relation to the appended drawings, in which: [Fig. 1] Figure 1, already described in relation to the prior art, presents a flowchart of a method for managing a magnetic signature of a ship according to an embodiment of the prior art; [Fig. 2] Figure 2 presents a flowchart of a method for managing a magnetic signature of a ship according to a particular embodiment of the invention; and [Fig.3] Figure 3 represents an example of the structure of a device for managing a magnetic signature of a ship according to an embodiment of the invention, allowing the implementation of the steps of the method of Figure 2. 5. DETAILED DESCRIPTION In all the figures of this document, identical elements and steps are designated by the same numerical reference. We now present, in relation to Figure 2, a method for managing a magnetic signature of a ship according to a particular embodiment of the invention. This method is for example implemented by a device 30, as described below in relation to Figure 3. In a particular embodiment, the device 30 is for example integrated into a portable measuring station on board the ship.In a step 21, the device 30 obtains magnetic field measurements each associated with information comprising a geographic position, a magnetic heading of the ship and a distance from the ship. As already mentioned above, the magnetic heading information of the ship can be obtained either directly (for example with a magnetometer), or indirectly (for example by calculation from the compass heading or the geographic heading). In a step 22, the device determines a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information. As detailed below, the resolution of the first inverse problem is carried out with a first resolution method based on a sparse regularization using an a priori favoring sparseness.Steps 21 and 22 are part of a step 20 of magnetic modeling of the ship, at the end of which a magnetic moment model of the ship is obtained, at a magnetic heading. As indicated by the frame referenced 20, steps 21 and 22 are iterated for each of a plurality of N magnetic headings (generally the following four (N = 4) particular magnetic headings are considered: north magnetic heading (when the magnetic heading is equal to 0°), south magnetic heading (when the magnetic heading is equal to 180°), east magnetic heading (when the magnetic heading is equal to 90°) and west magnetic heading (when the magnetic heading is equal to 270°)). A magnetic moment model of the ship is thus obtained for each of this plurality of magnetic headings.In a step 23, the device performs a prediction of the magnetic signature of the ship at a geographical position and / or a magnetic heading of the ship and / or a distance from the ship other than that or those of the magnetic field measurements, as a function of the magnetic moment model of the ship (magnetic moment domain) determined in step 22. In a step 24, the device performs a compensation of the magnetic signature of the ship using an immunization system (degaussing system) comprising current loops installed on board the ship.The compensation step 24 itself comprises the following steps: - determination in step 25 of a magnetic moment(s) model of each of the current loops, by simple subtraction between the magnetic moment model obtained for the ship with an activated loop and the magnetic moment model obtained for the ship without an activated loop; - for each of the current loops, determination in step 26 of a compensation current as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops; and - application in step 27, to the current loops, of the determined compensation currents.As detailed below, in a particular implementation, the calculation of the compensation currents is carried out by solving a second inverse problem, aiming to determine the compensation currents as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops, with a second resolution method based on a sparse regularization using a priori favoring sparseness. We now present in more detail various principles and characteristics allowing to explain the process of managing a magnetic signature of a ship in Figure 2 and the various steps included in it. General principle (magnetic modeling of the ship, prediction and compensation) The ship is considered as a plane of dipoles / magnetic moments. Each point of this plane is thus decomposed into three axes and the set of magnetic moments associated with certain points of this plane are noted as follows: - m.^ : vector of moments along the longitudinal axis of the ship (we will consider a positive moment if it is oriented forward); - m ^ : vector of moments along the transverse axis of the ship (we will consider a positive moment if it is oriented towards starboard); and - m ^: vector of moments along the vertical axis of the ship (we will consider a positive moment if it is oriented downwards). The magnetic signature of the ship in the static domain is mainly due to two distinct contributions: - permanent magnetization: linked to the intrinsic magnetic characteristics of the ferromagnetic materials constituting the ship and to their history (remanence). This contribution is considered to vary slowly except under certain specific conditions. For signature prediction and degaussing adjustment, this contribution is considered constant; and - induced magnetization: linked to the presence of ferromagnetic materials in the Earth's field. This contribution is therefore variable depending on the location of the ship (local Earth's field) and the attitude of the ship (magnetic heading, roll, pitch). We thus define: m ^ = m ^^^^^ ± m ^^^^ om ^^^^^: vector of permanent longitudinal moments om ^^^^ : vector of induced longitudinal moments m ^ = m ^^^^^ ± m ^^^^ om ^^^^^ : vector of permanent transverse moments om ^^^^ : vector of induced transverse moments m ^ = m ^^^^^ ± m ^^^^ om ^^^^^ : vector of permanent vertical moments om ^^^^: vector of induced vertical moments. The magnetic modeling of the ship will therefore consist of estimating the moment vectors (longitudinal, transverse and vertical) and separating the two contributions (permanent and induced). To do this, in a conventional system, the ship passes over an underwater measuring station where tri-axial vector sensors are aligned and fixed (fixed measurement depth) and where the ship passes above this line so that some sensors are on the port side and others on the starboard side. The trace of the magnetic signature on each of the sensors is thus recovered via the three components of the measured magnetic field (Bx, By, Bz).This measurement is carried out at each of the four specific magnetic headings mentioned above (north magnetic heading, south magnetic heading, east magnetic heading and west magnetic heading) so that the local Earth's magnetic field is, for its horizontal part, aligned with one of the horizontal axes of the ship (longitudinal or transverse) in order to be able to model the ship from these components of the magnetic field: - North magnetic heading: the permanent part is always present and the Earth's field will induce a magnetization in the vertical axis and the longitudinal axis: ^ = m. ^^^^^ + m ^^^^ + m ^^^^^ + m ^^^^^ + m ^^^^ ; - South magnetic heading: the permanent part is always present and the Earth's field will induce magnetization in the vertical axis and the longitudinal axis: ^ = m ^^^^^ − m ^^^^ + m ^^^^^ + m ^^^^^ + m ^^^^; - West magnetic heading: the permanent part is always present and the Earth's field will induce magnetization in the vertical axis and the transverse axis: ^ = m ^^^^^ + m ^^^^^ + m ^^^^ + m ^^^^^ + m ^^^^ ; and - Magnetic heading is: the permanent part is always present and the Earth's field will induce magnetization in the vertical axis and the transverse axis: ^ = m ^^^^^ + m ^^^^^ − m ^^^^ + m ^^^^^ + m ^^^^ . Without a compensation loop of the vertical earth field, one cannot eliminate m ^^^^. By obtaining the different magnetic moments of the ship, it is thus possible to recalculate the magnetic signature of the ship according to another terrestrial field (another geographical position) and / or at a different orientation (another magnetic heading). On the other hand, the change in distance requires having a model close to physical reality (which the invention allows, unlike the state of the art) to have validity outside the initial measurement zone. For the degaussing system, the principle is based on current loops positioned along one of the three axes of the ship: - Loop L: loop creating a longitudinal magnetic moment; - Loop A: loop creating a transverse magnetic moment; and - Loop M: loop creating a vertical magnetic moment. The number of loops of each type (L, A and M) and their characteristics depend on the desired level of magnetic signature reduction (compensation).To carry out the degaussing adjustment, we will therefore, classically, return to the measuring station at a magnetic heading as many times as we have loops in order to measure the magnetic effect of each of the loops at a known current Iref. We will thus obtain in the measurement the magnetic effect of the ship already measured alone and the magnetic field created by the active loop. Schematically, a conventional degaussing system therefore seeks to estimate the currents from magnetic field measurements acquired by a measuring station in the three axes Bx, By, Bz for the four particular magnetic headings mentioned above (north magnetic heading, south magnetic heading, east magnetic heading and west magnetic heading). In a particular implementation, the present invention aims to adjust the currents directly in the space of magnetic moments instead of reasoning in magnetic field.Operating principle of a degaussing system As indicated above, the degaussing system consists of loops L, A and M. Let us consider for example four loops of each type. The adjustment therefore consists of finding the currents to be injected into each of the loops by dividing this current into two parts: - I. ^^^^^ : current in loop X which is constant to compensate for permanent magnetization; and - I ^^^^ : current in the X loop which is variable to compensate for an induced magnetization. The adjustment is made at a measuring station to a given local terrestrial field. The system will thus calculate, by rule of proportionality, the induced current to be injected for a different terrestrial field and attitude (projection of the terrestrial field in the ship's frame). The principle of degaussing is therefore to find: I ^^^^^ , I ^^^^^ , I ^^^^^ , I ^^^^^ allowing to compensate m ^^^^ ; I ^^^^^^ , I ^^^^^^ , I ^^^^^^ , I^^^^^^ allowing to compensate m ^^^^^ ; I ^^^^^ , I ^^^^^ , I ^^^^^ , I ^^^^^ allowing to compensate m ^^^^ ; I ^^^^^^ , I ^^^^^^ , I ^^^^^^ , I ^^^^^^ allowing to compensate m ^^^^^ ; I ^^^^^ , I ^^^^^ , I ^^^^^ , I ^^^^^ allowing to compensate m ^^^^ ; and I ^^^^^^ , I ^^^^^^ , I ^^^^^^ , I ^^^^^^ allowing to compensate m ^^^^^ . Definition of the problem for compensation (degaussing system) We are looking for the magnetic model of the ship in magnetic moments: m ^^^^ and m ^^^^^ ; m ^^^^ and m ^^^^^ ; and m ^ = m ^^^^^ ± m ^^^^(not separable unless there is a compensation loop for the vertical Earth field at the measuring station). This model is calculated from the magnetic field measurements (Bx, By, Bz) at the four particular magnetic headings mentioned above on X sensors. For the degaussing system, the measurement of each loop effect is carried out at a given current (here ^ ^^^ which is the same regardless of the loop, in practice it can be different). We model each loop X by a moment ^ ^ (here only one but we can find several, the important thing being that a longitudinal loop (respectively transverse or vertical) compensates a longitudinal moment (respectively transverse or vertical) of the ship. We are looking for currents such as: ^ ^ ^ ^ 1 With U ^ = ^ ⋮ ^ of dimension x*1 and l, t, v respectively the number of moments in the vector 1 m ^ , m ^ and m ^ respectively. Noting: ) ^ ^ ^ ^ ⋮ ⋮ ⋮ 0 0 0 0 0 0 0 0 0 0 0 0 We obtain in matrix form: ^ ^^^^^ ^ Ship's magnetic field measurements ^ The measurement ^ ^^^^^^ is worth: with the coefficient vectors kn depending on the position of the measurement. These coefficients therefore have a different value for each measurement point. Measurements of loop effects The measurement will contain the magnetic effects of the ship and the activated loop. Since the magnetic field is governed by the principle of superposition, we can, knowing the measurement of the magnetic effect of the ship, deduce the effect of the loop alone. measured are worth: ^ ^ ^ ^^^ ^8 ^ ^^^ ^3 ^ magnetic fields measured, loop effects and currents sought According to the equations and assuming a perfect system allowing the same measurement points to be made for the ship and the loop effects (or by modeling and recalculation at the same point), we obtain: ^^^^^^ ^ ^ ^^^^^ ^ These equations clearly show the advantage of the present invention in being able to solve the problem directly in the domain of magnetic moments (equation 6) rather than in that of magnetic fields (equation 11) as in classical methods. On the one hand, we are freed from geometry (kn terms) and on the other hand, the problem to be solved is of lower dimension. Properties of moments at magnetic headings As described above, the measurement is carried out at four particular magnetic headings mentioned above, so that: - North magnetic heading: ^ ^ = ^ ^^ + ^ ^^ + ^ ^^ with: ^ ^^ = m ^^^^^ + m ^^^^ ^ ^^ = m ^^^^^ ^ ^^ = m ^^^^^ + m ^^^^ = ^ ^ - South magnetic heading: ^ ^ = ^ ^^ + ^ ^^ + ^ ^^ with: ^ ^^ = m ^^^^^ − m ^^^^ ^ ^^ = m ^^^^^ ^ ^^= m ^^^^^ + m ^^^^ = ^ ^ - West magnetic heading: ^ ^ = ^ ^^ + ^ ^^ + ^ ^^ with: ^ ^^ = m ^^^^^ ^ ^^ = m ^^^^^ + m ^^^^ ^ ^^ = m ^^^^^ + m ^^^^ = ^ ^ - : ^ ^ ^ ^ ^ = ^ ^^ = ^ ^^ = ^ ^^ = ^ ^^ (16) Magnetic fields measured for the passes at the four particular magnetic headings (north magnetic heading, south magnetic heading, east magnetic heading and west magnetic heading) The measurement giving the magnetic field, we can therefore write that: ^ ^ ^ ^ We will thus, by modeling, and with equations (12) to (16), find ^ ^^^^ , ^ ^^^^^ , ^ ^^^^ , ^ ^^^^^ and ^ ^ (step 22 of figure 2). Measured loop effects Each loop is measured at any magnetic heading with a known current (here ^ ^^^) since the magnetic moment, and therefore the magnetic field, created by a constant current loop does not depend on the magnetic heading. Whereas previous methods removed from the measurement the magnetic field created by the ship alone at the same magnetic heading to obtain the magnetic field created by a loop according to the following equations: ^ ^^^ ^1 ^ ^^^ ^2 ^ ^^^ ^3 ^ proposed method will measurements and determine the magnetic moments created by the loops by difference between models in magnetic moments at the same magnetic heading (we subtract from the model resulting from the measurement including the ship and an active loop, the model resulting from the measurement of the ship alone at the same magnetic heading). We will thus, by modeling, find ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ , ^ ^^ and ^ ^^(step 25 of figure 2). Adjusting the currents We are looking to compensate for the magnetic effect of the ship with the degaussing loops. The problem is therefore broken down as follows: - we are looking for the linear combination of the longitudinal loops which compensates for the induced longitudinal moment(s) of the ship: - we are looking for the linear combination of longitudinal loops which compensates for the permanent longitudinal moment(s) of the ship: - we seek the linear combination of transverse loops which compensates the induced transverse moment(s) of the ship: - we are looking for the linear combination of transverse loops which compensates the transverse moment(s) of the ship: ^ ^^^ - we are looking for the linear combination of vertical loops that compensates the − ^ ^^^The same type of calculation as for the other axes is possible if an Earth field compensation loop is used to measure the ship. Without this, the vertical permanent induced adjustment is achieved by experiment. Hence the problem to be solved: ^ ^^^^ = ^ ^^ ^^ ^^ = − ^ ^ + ^ ^^^^^ ^ ^^ + ^ ^^^^^ ^ ^^ + ^ These equations are a simplified form considering the loops as perfect and equivalent to a single moment at the measurement distance. These equations can be made more complex depending on the reality of the models obtained (multiple moments in different directions). In this case, the impact of a loop in the other axes must be taken into account (minimal in principle, except for particular geometry). More generally (written in generic form), we seek to compensate the ship with p+q+r degaussing loops (p loop(s) of type L, q loop(s) of type A and r loop(s) of type M). The problem is therefore broken down as follows: - we seek the linear combination of the longitudinal loops which compensates the induced longitudinal moment(s) of the ship: ^ ^^^^ - we are looking for the linear combination of longitudinal loops which compensates for the permanent longitudinal moment(s) of the ship: ^ ^^^^^ = − - we are looking for the linear combination of the loops the induced transverse moment(s) of the ship: ^ ^^^^ = ^ ^^^ - we are looking for the linear combination of transverse loops which compensates for the permanent(s) of the ship: ^ ^^^^^ = − ^ ^^^ ^^^ - we are looking for the combination that compensates for the vertical moment(s) of the vessel: ^ = − ^ ^^^ ^^^ The same type of calculation as for the other axes is possible if an Earth field compensation loop is used to measure the ship. Without this, the vertical permanent induced adjustment is achieved by experiment. Hence the problem to be solved: ^ ^ ^ ^ ^ ^ These equations are a simplified form considering the loops as perfect and equivalent to a single moment at the measurement distance. These equations can be made more complex depending on the reality of the models obtained (multiple moments in different directions). In this case, the impact of a loop in the other axes (minimal in principle) must be taken into account. The problem considered through equations (26') to (30') is written in the following matrix form: ^ ^^^^^^ ^ = − ^ ^^^ ^ ^^^^^^^ ^ (31) ^ ^^^^^^ = ^ ^ ^ ^ ^ ^ ^ loops of dimension (l+t+v)*(p+q+r), and ^ the vector of currents to be estimated. Estimation of the magnetic moments of the ship from the measured magnetic fields (step 22 of figure 2) magnetic fields of the ship (i.e. the determination of the ship's magnetic moment model) from the magnetic fields is a well-known inverse problem in ship magnetism. The angle taken in this invention is to make the hypothesis (which is verified in practice) that the sources of magnetization (or dipole moments) which explain the magnetic fields observed outside the near-field zone are limited in number (the nature of the physical sources of magnetization are known to sailors and this principle is consistent with the design of the degaussing system with a limited number of loops and therefore equivalent magnetic moments). Mathematically, this type of inverse problem finds a relevant translation in the way of regularizing the problem; regularizing amounts to imposing constraints on the expected solutions.A commonly used state-of-the-art regularization of the problem for inverse problems is that of Tikhonov (also called norm. , noted ‖ ^ ‖ ^ and which contains the sum of the squares of the regularized variable), such that: ^ with ^ ^^^^^^ the vector of magnetic moments of the model into magnetic moments of the ship, ^ ^^^^^^ the vector of the ship's magnetic field measurements, and the matrix ^ (matrix of position-dependent coefficients associated with the magnetic field measurements) such that ^ = with [^] ^^ = ^ x,y,z being the coordinates in the ship-reference frame and r=sqrt(x²+y²+z²). This (and its variants) is however not relevant with regard to physical reality because it does not allow to take into account that the magnetization sources are localized and in limited number. As already mentioned above, the present invention is based on the use of a resolution method based on a sparse regularization (using a priori favoring sparseness, also called "sparse a priori" or "sparse promoting priors" in English) to solve the inverse problem aiming to determine the magnetic moments of the ship from the magnetic field measurements and the associated information. The use of such a sparse regularization method in step 22 is justified by physical reality: it is thus possible to automatically estimate a limited number of sources explaining the measured magnetic fields.In embodiments of the invention of step 22, the sparse regularization-based solution method relies on at least one of the norms. ^ . So, inverse problem can be written in several mathematical forms, as listed below. For the norm ℓ ^ : For the standard Or equivalently: It is also possible to combine the penalties of the standards ^ and ℓ ^ : ^ Or to combine the penalties of the standards ℓ, ^ and ℓ ^ ^ Many algorithms exist to solve these problems, which are found in many scientific fields. A compromise in practice must be found according to the available computing power, the ease of implementation, the expected performance, the estimation of the uncertainty on the vector ^ ^^^^^^. Two large families of algorithms can be called upon for these problems: algebraic optimization algorithms on the one hand and simulation algorithms on the other. In a particular embodiment, for the estimation of the ship's magnetic moments, a Bayesian method and the associated algorithm give a good compromise between convergence speed and quality of estimation of the vector ^ ^^^^^^, while providing the uncertainties for the estimated moments. Estimation of currents from the ship's magnetic moments (step 26 of figure 2) The ship's magnetic moments (dipole moments) having been estimated (step 22 of figure 2), it is then a matter, in step 26 of figure 2, of estimating the compensation currents from these. The same parsimony hypothesis of the solution to the set of equations (26) to (30) - or more generally (26') to (30') - is adopted and an a priori favoring parsimony is chosen to find this solution. In other words, certain currents may be zero in a given situation depending on the magnetic moments estimated in step 22. In embodiments of the invention of step 22, the resolution method based on a parsimonious regularization relies on at least one of the standards Thus, the inverse problem can be written in several mathematical forms, as listed below. For the norm ℓ ^ : ^ ^^^ ∥ ^ ∥ ^ under duress with ^ the vector of compensation currents to be estimated, ^ ^^^^^^^ the loop moment matrix (i.e., a matrix of coefficients depending on the magnetic moment models of the current loops) and ^ ^^^ a reference current (already discussed above). An example of the vector ^ is mentioned above in equation (6). For the standard ^ ^^^^ 1 ^^^ ∥ ^ ∥ under ^ ^^^^^^^ ^ ^ the constraint ^ = − ^ ^ ^ ^^^ Or equivalently: ^ ^^^ It is also possible to combine the penalties of the ℓ standards ^ and ℓ ^ : ^ ^ ^^^ Or again ^ ^ ^ ^ ^ ^ ^ ^^^For this step 26 of the method, the resolution method based on sparse regularization uses, for example, a Markov Chain Monte Carlo type algorithm. In other words, the estimation of the vector ^ is for example ensured by an MCMC type algorithm. But any other approach is conceivable. We now present, in relation to Figure 3, an example of the structure of a device 30 for managing a magnetic signature of a ship according to an embodiment of the invention, allowing the implementation of the steps of the method of Figure 2. The device 30 comprises a random access memory 33 (for example a RAM memory), a processing unit 31 equipped for example with a processor, and controlled by a computer program 320 stored in a read-only memory 32 (for example a ROM memory or a hard disk).Upon initialization, the code instructions of the computer program are for example loaded into the RAM 33 before being executed by the processor of the processing unit 31. This figure 3 illustrates only one particular way, among several possible ones, of producing the device 30 so that it performs certain steps of the method for managing a magnetic signature of a ship according to the invention (according to any one of the embodiments and / or variants described above in relation to figure 2). Indeed, these steps can be carried out indifferently on a reprogrammable computing machine (a PC computer, a DSP processor or a microcontroller) executing a program comprising a sequence of instructions, or on a dedicated computing machine (for example a set of logic gates such as an FPGA or an ASIC, or any other hardware module).In the case where the device 30 is produced with a reprogrammable computing machine, the corresponding program (i.e. the sequence of instructions) may be stored in a removable storage medium (such as for example a CD-ROM, a DVD-ROM, a USB key) or not, this storage medium being partially or totally readable by a computer or a processor.

Claims

CLAIMS 1. Method for managing a magnetic signature of a ship, executed by a computing machine and comprising a step (20) of magnetic modeling of the ship itself comprising: - obtaining (21) magnetic field measurements each associated with information comprising a geographical position, a magnetic heading of the ship and a distance from the ship; and - determining (22) a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information; characterized in that the resolution of the first inverse problem is carried out with a first resolution method based on a sparse regularization using an a priori favoring sparseness. 2.The method of claim 1, wherein the first sparse regularization-based resolution method relies on at least one of the ℓ norms. ^ 3. Method according to claim 2, in which the first inverse problem is written in one of the following mathematical forms: ^ ^ ^ ^ ^^^^^^ ∥^ ^ with ^ ^^^^^^ a vector of the ship's magnetic moments to be estimated, ^ ^^^^^^ a vector of magnetic field measurements and ^ a matrix of position-dependent coefficients associated with the magnetic field measurements.

4. A method according to any one of claims 1 to 3, wherein the first solution method based on sparse regularization is a Bayesian method.

5. Method according to any one of claims 1 to 4, comprising a step (23) of predicting the magnetic signature of the ship at a geographical position and / or a magnetic heading of the ship and / or a distance from the ship other than that or those of the magnetic field measurements, as a function of the magnetic moment model of the ship. 6.Method according to any one of claims 1 to 4, comprising a step (24) of compensating the magnetic signature of the ship using an immunization system comprising current loops installed on board the ship, the step of compensating the magnetic signature of the ship comprising: - determining (25) a magnetic moment(s) model of each of the current loops; - for each of the current loops, determining (26) a compensation current as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops; and - applying (27), to the current loops, the determined compensation currents. 7.Method according to claim 6, in which the calculation of the compensation currents is carried out by solving a second inverse problem, aiming to determine the compensation currents as a function of the magnetic moment model of the ship and the magnetic moment models of the current loops, with a second resolution method based on a sparse regularization using an a priori favoring sparseness.

8. Method according to claim 7, in which the second resolution method based on a sparse regularization relies on at least one of the ℓ norms. ^ 9. The method of claim 8, wherein the second inverse problem is written as one of ^ ^ ^ ^ ^ ^^^ ^^^ ^ ^^^^^^ 1 ^ ^^^^^^^ ^ ^ ^ ^ ^ ^ ^ ​a matrix of coefficients depending on the models in magnetic moments of the current loops and ^ ^^^a reference current.

10. Method according to any one of claims 7 to 9, in which the second solution method based on a sparse regularization uses a Markov Chain Monte Carlo type algorithm.

11. Computer program product (320), comprising program code instructions for implementing the method according to at least one of claims 1 to 10, when said program is executed on a computer.

12. Computer-readable and non-transitory storage medium (32), storing a computer program product (320) according to claim 11. 13.Device (30) for managing a magnetic signature of a ship, the device comprising a dedicated or reprogrammable computing machine (31, 32, 33), configured to perform a step of magnetic modeling of the ship itself comprising: - obtaining magnetic field measurements each associated with information comprising a geographical position, a magnetic heading of the ship and a distance from the ship; and - determining a magnetic moment model of the ship, by solving a first inverse problem aimed at determining magnetic moments of the ship from the magnetic field measurements and the associated information; characterized in that the resolution of the first inverse problem is carried out with a first resolution method based on a sparse regularization using an a priori favoring sparseness.

14. Device according to claim 13, in which the computing machine is also configured to perform a step of predicting the magnetic signature of the ship at a geographical position and / or a magnetic heading of the ship and / or a distance from the ship other than that or those of the magnetic field measurements, as a function of the magnetic moment model of the ship. 15.Device according to claim 13, wherein the computing machine is also configured to perform a step of compensating the magnetic signature of the ship using an immunization system comprising current loops installed on board the ship, the step of compensating the magnetic signature of the ship comprising: - determining a model in magnetic moment(s) of each of the current loops; - for each of the current loops, determining a compensation current as a function of the model in magnetic moments of the ship and the models in magnetic moments of the current loops; and - applying, to the current loops, the determined compensation currents.

Images