Device and method for estimating an interferometric phase shift, electronic device for characterizing a fluid

Abstract

This device (22) for estimating a phase shift, between two input signals (s1, s2) originating from the same periodic signal (s0) of period T processed by interferometry, comprises a multimodal coupler (24) receiving the two input signals (s1, s2) and providing a multiphase system (S), a unit (30) for processing the multiphase system (S) for calibration and estimation (SG) of the phase shift, and means (32) for storing calibration parameters. The processing unit (30) is configured (34) to test whether the multiphase system (S) meets predetermined conditions required to carry out an estimation of the calibration parameters and to: carry out the estimation of the calibration parameters and their recording in the storage means (32) if the predetermined conditions required are met; extracting, from the storage means (32), the calibration parameters of a previous estimate if the required predetermined conditions are not met.Figure for abstract: Fig. 1.

Description

Title of the invention: Device and method for estimating an interferometric phase shift, electronic device for characterizing a fluid

[0001] The present invention relates to a device for estimating a phase shift between two input signals originating from the same periodic signal processed by interferometry, an electronic device for characterizing a fluid with a transducer comprising such a phase shift estimation device, and a method for estimating a phase shift between two input signals originating from the same periodic signal processed by interferometry.

[0002] An electronic device for characterizing a fluid, liquid or gaseous, by interferometry comprises at least one sensor designed to interact with the fluid and a transducer processing, by interferometry, the signals resulting from this interaction to provide a sequence of at least one measurement signal representative of an interaction of the fluid with each sensor.

[0003] It is understood that the interaction with the fluid, the provision of the sequence of measurement signals and its possible processing are operations which can be carried out in parallel over time. Alternatively, the processing can be carried out a posteriori.

[0004] This type of device is sometimes called a multivariate sensor, particularly when it comprises a plurality of sensors. It can be considered for various olfactory estimation applications and is generally, in this case, referred to as an “electronic nose” (in a gaseous medium) or “electronic tongue” (in a liquid medium). It is then used to detect, discriminate, identify and quantify volatile organic compounds in a gaseous fluid to be analyzed, or compounds present in a liquid. It can be used in various industrial fields such as: - the perfume industry, for example to compare, study or design pure or mixed olfactory compositions, - environmental protection, in particular to detect odorous pollution or monitor the quality of more or less confined environments, - monitoring of industrial sites presenting a risk of contamination by volatile or solvent-soluble materials that are potentially dangerous or odorous, - health, for example to offer a smell substitute to people suffering from anosmia or to detect volatile biological markers such as the emanations of infectious microbiological activity, - the food industry, for example to detect contamination in a food manufacturing and / or distribution chain, - any other industrial field in which the control of any odorous product may prove useful.

[0005] The invention applies more particularly to a device for estimating a phase shift between two input signals originating from the same periodic signal of period T processed by interferometry, comprising: - a multimodal coupler with two inputs, for receiving the two input signals, and with N outputs, N > 3, for providing a multiphase system of N signals formed to be phase-shifted by T / N between them; and - a multiphase system processing unit to extract calibration parameters and an estimate of the phase shift between the two input signals.

[0006] Such a phase shift estimation device is particularly advantageous for an electronic device for characterizing a fluid by interferometry because it makes it possible to determine without any ambiguity modulo T the evolution over time of the phase shift, characteristic of the interaction of the fluid with any suitable sensor, between the two signals supplied to it. But it is more generally applicable to any system producing two signals, originating from the same periodic signal of period T processed by interferometry, for which the phase shift is to be determined at any instant without ambiguity modulo T. In general, the periodic signal is of sinusoidal form and of period expressed angularly, hence a period T generally equal to 2ir.

[0007] It is known to apply a Clarke transformation (or any other equivalent transformation such as a Concordia or Fortescue transformation) to the multiphase system to theoretically obtain a centered circular representation from which it is then very easy to extract a value of the phase shift to be estimated at each instant by simple application of the Arc tangent function noted Arctan in the case of sinusoidal signals whose periodicity is equal to 2ir when it is expressed angularly. It then often remains necessary to carry out a phase unwrapping operation to deal with the modulo discontinuities.

[0008] But in reality, the multiphase system presents defects which are due to imperfections in the interferometry processing, the multimodal coupler, and a possible sensor at the output of the multimodal coupler (for example a photographic sensor CCD when it comes to optical interferometry) and possibly the periodic signal generator from which the two input signals come. These defects appear in particular in the form of differences between the amplitude values of the N signals that constitute it: by multimodal coupling and possible optical capture, these N signals should have the same amplitude, but this is not the case. The observed defects also generally appear in the form of amplitude shifts of the N signals relative to the null axis around which they should evolve: by generation of the initial periodic signal, multimodal coupling and possible optical capture, these N signals should all evolve around the null axis, or at least with the same shift relative to this null axis, but this is not the case either. As a result, the Clarke transformation of such an imperfect multiphase system provides on a sequence, not a centered circle, but an off-centered ellipse.Thus, an application of the Arctan function to the result of the Clarke transformation at each instant produces a biased estimate of the phase shift.

[0009] A first solution to resolve these defects of the multiphase system is provided in the article by Halir et al, entitled "Direct and sensitive phase readout for integrated waveguide sensors", published in IEEE Photonics Journal, volume 5, number 4, in August 2013. It consists of estimating calibration parameters by coordinate change from a recalibration of the ellipse obtained by Clarke transformation on a centered circle, on a sufficiently long sequence, to record this recalibration in memory as calibration parameters, then to apply this recalibration by coordinate change at each instant on the Clarke transformation of the multiphase system, before carrying out the phase unrolling. This approach is advantageous because the calibration only requires the data of the multiphase system itself, which allows it to be described as blind calibration.On the other hand, the estimation of blind calibration parameters can be very imprecise, especially when the measurements used for calibration do not allow the ellipse from which the calibration parameters are calculated to be reconstructed with a certain precision. However, nothing is provided to ensure this and taking measurements over a sufficiently long sequence does not guarantee anything.

[0010] A second solution to resolve these defects of the multiphase system, which takes the form of an improvement relatively close to the first, is provided by patent document WO 2022 / 238170 A1. It consists of estimating the calibration parameters from the amplitude and amplitude shift values of the multiphase system over a sufficiently long sequence, recording these parameters in memory, then applying this calibration by processing at each instant of the multiphase system before applying the Clarke transformation and the phase unrolling operation to it. This again advantageously involves a blind calibration. But this second solution has the same drawback as the previous one, namely an estimation of the calibration parameters which can prove to be very imprecise, in particular when the measurements used for calibration do not allow the amplitude and amplitude shift values of the multiphase system to be evaluated. However, nothing is provided to ensure this and taking measurements over a sufficiently long sequence does not guarantee anything in this case either. However, it has the merit of showing that it is possible to carry out blind calibration directly from the amplitude and amplitude shift values of the multiphase system.

[0011] It may thus be desirable to provide a phase shift estimation device with blind calibration which makes it possible to overcome at least some of the aforementioned problems and constraints.

[0012] A device is therefore proposed for estimating a phase shift between two input signals originating from the same periodic signal of period T processed by interferometry, comprising: - a multimodal coupler with two inputs, for receiving the two input signals, and with N outputs, N > 3, for providing a multiphase system of N signals formed to be phase-shifted by T / N between them; - a multiphase system processing unit to extract calibration parameters and an estimate of the phase shift between the two input signals; and - means of storing calibration parameters; wherein the processing unit is configured to test whether the N signals of the multiphase system meet predetermined conditions required to perform an estimation of the calibration parameters and to: - carry out the estimation of the calibration parameters and their recording in the storage means if the required predetermined conditions are met; and - extract from storage means the calibration parameters of a previous estimate if the required predetermined conditions are not met.

[0013] Thus, by providing a test, whatever it may be, on predetermined conditions identified as being required to carry out the estimation of the blind calibration parameters, and also by providing for reusing the calibration parameters of a previous estimation which fulfilled the conditions of the test when the latter is not favorable for a new calibration, an improvement of the latter is noted. This proposal is both simple and clever. It is then within the reach of the person skilled in the art to identify which are the predetermined conditions to apply to carry out the test according to the proposed calibration. For example, if the proposed calibration is that of the aforementioned Halir et al document, these may relate to the possibility of carrying out elliptical recalibration, while if it is that of the patent document WO 2022 / 238170 A1, they should rather relate to the ability to find the amplitude and amplitude offset values of the multiphase system. Other predetermined conditions can be adapted to other conceivable calibration methods.

[0014] Optionally, the predetermined required conditions that the processing unit is configured to test include that each signal of the provided multiphase system results from enough different phase shift values so that they cover at least the period T.

[0015] Also optionally, the test carried out by the processing unit includes a verification that each signal of the supplied multiphase system varies sufficiently in amplitude to deduce its extreme values which can be reached.

[0016] Also optionally, the predetermined required conditions that the processing unit is configured to test include that a Clarke, Concordia or Fortescue transformation of the provided multi-phase system produces enough different phase shift values to define a unique elliptic fit ellipse solution.

[0017] Also optionally, the processing unit is provided with means for filtering the multiphase system designed to estimate a state of this multiphase system from a measurement, these filtering means being more precisely defined by the fact that: - the state to be estimated comprises constant values of amplitudes and amplitude shifts of the N signals of the multiphase system, a variable value relating to the phase shift to be estimated and a constant value to aid estimation of the variable value; and - the measurement includes the N signals of the multiphase system as provided by the multimodal coupler; and the calibration parameters include the constant values of amplitudes and amplitude shifts of the N signals of the multiphase system.

[0018] Also optionally, the processing unit is configured to: - perform a Clarke, Concordia or Fortescue transformation of the multiphase system and deduce parameters of an ellipse fitted to this Clarke, Concordia or Fortescue transformation; - carry out a calibration of this adjusted ellipse by parametric transformation into a recentered circle, in which the calibration parameters include the parameters of the transformation; then - estimate the phase shift between the two input signals after applying the calibration parameters to the Clarke transformation of the multiphase system.

[0019] Also optionally, the processing unit is configured to: - carrying out a calibration of the N periodic signals of the multiphase system from amplitude values and amplitude shifts of these N periodic signals, in which the calibration parameters comprise a combination of these amplitude values and amplitude shifts; - perform a Clarke, Concordia or Fortescue transformation of the multiphase system after calibration and deduce parameters of a centered circle adjusted on this Clarke, Concordia or Fortescue transformation; then - estimate the phase shift between the two input signals from this Clarke, Concordia or Fortescue transformation.

[0020] An electronic device for characterizing a fluid is also proposed, comprising: - at least one sensor designed to interact with the fluid; and - a transducer designed to provide, in interaction with said at least one sensor, a sequence of at least one measurement signal representative of an interaction of the fluid with each sensor; in which the transducer is a system for processing a periodic signal by interferometry and the electronic characterization device comprises at least one device for estimating a phase shift between two input signals from said periodic signal as defined previously.

[0021] Optionally, the transducer is a system for amplifying optical index variation by Mach-Zehnder interferometry with multimodal interference, the multimodal coupler then having optical outputs and associated with an optical sensor.

[0022] A method is also proposed for estimating a phase shift between two input signals originating from the same periodic signal of period T processed by interferometry, comprising: - a multimodal coupling of the two input signals for the provision of a multiphase system of N signals formed to be phase-shifted from each other by T / N where N > 3; - processing of the multiphase system to extract calibration parameters and an estimation of the phase shift between the two input signals; and - a recording of the calibration parameters in storage means; wherein the processing comprises a test on predetermined conditions required to perform an estimation of the calibration parameters and is designed to selectively perform one or other of the following operations: - the estimation of the calibration parameters and their recording in the storage means if the predetermined conditions required are met by the N signals of the multiphase system; and - the extraction, from storage means, of calibration parameters of a previous estimate if the predetermined conditions required are not met by the N signals of the multiphase system.

[0023] The invention will be better understood with the aid of the following description, given solely by way of example and with reference to the appended drawings in which: - [Fig.l] schematically represents the general structure of an interferometric system comprising a phase shift estimation device according to an embodiment of the invention, - [Fig.2] schematically represents the general structure of an interferometric system comprising a shared plurality of phase shift estimation devices such as that of [Fig.l], - [Fig.3] illustrates the successive steps of an interferometric method including a phase shift estimation according to an embodiment of the invention, - [Fig.4] shows the successive stages of the interferometric process of [Fig.3] according to a first variant embodiment, - [Fig.5] shows the successive stages of the interferometric process of [Fig.3] according to a second variant embodiment, - [Fig.6] schematically represents the general structure of an electronic device for characterizing a fluid according to one embodiment of the invention, - [Fig.7A] illustrates an example of a time diagram in which several response signals, or sensorgrams, which can be obtained by the electronic device of [Fig.6], are superimposed, under controlled multivariate measurement conditions, - [Fig.7B] illustrates, in the form of a circular diagram, an example of a standardized signature that can be calculated by the electronic device of [Fig.6] from sensorgrams such as those of [Fig.7A], - [Fig.8] illustrates the successive steps of a method for characterizing a fluid that can be implemented by the electronic device of [Fig.6], and - [Fig.9] schematically illustrates the improvement provided by an interferometric method including a blind calibration phase shift estimation according to the invention, in particular according to the first variant embodiment of [Fig.4],

[0024] The interferometric system 10 shown schematically in [Fig.l] comprises an interferometer 12, for example a Mach-Zehnder optical interferometer. Such an interferometer 12 receives a single periodic coherent optical signal s0 of period T, in particular a sinusoidal optical signal s0 of pulsation 2irf where f is the carrier frequency, generated and emitted, in a waveguide, for example by a laser source 14 of coherent light. The period T can be defined temporally, in which case it is the inverse of the carrier frequency, or angularly, in which case it is 2ir for a sinusoidal signal.

[0025] In the interferometer 12, the periodic signal s0 is conducted to two distinct optical branches generally of the same length, one constituting a reference optical branch 16, the other an interaction optical branch 18. The first, the reference optical branch 16, produces a first periodic optical signal Si whose phase shift with respect to s0 depends on the length traveled in this reference optical branch 16. The second, the interaction optical branch 18, produces a second periodic optical signal s2 whose phase shift with respect to s0 depends not only on the same length traveled in this interaction optical branch 18, but also on an optical interaction with an interaction device 20 arranged on this path. The possible phase shift between Si and s2 therefore depends directly on the optical interaction with the device 20.

[0026] For a fluid, liquid or gaseous characterization application, the interaction device 20 is a sensor designed to interact with the fluid, including in particular a reactive site onto which are grafted reactive components capable of interacting by adsorption / desorption with compounds likely to be present in the fluid. By these interactions, the refractive index in the optical interaction branch 18 is locally modifiable and variable over time, which generates the aforementioned phase shift between Si and s2, as well as its variation over time which produces a temporal phase shift signal generally called a sensorgram.

[0027] Other applications are conceivable and any interaction device 20 capable of having an impact on the phase shift of the signal s2 is suitable. More generally, depending on the applications, other interferometric systems may be suitable, other than a Mach-Zehnder optical interferometer. It is for example possible to provide a Fizeau, Fabry-Pérot, Jamin, Ramsey-Bordé, Michelson, etc. interferometer, or even a combination of interferometers.

[0028] At the output of the interferometer 12, the two periodic signals Si and s2 could be recombined to interfere with each other and produce a single baseband signal carrying the phase shift information. But the latter is then ambiguous, so that they are advantageously provided as input signals of a device 22 for estimating their phase shift which will now be detailed.

[0029] This device 22 comprises a multimodal coupler 24 with two inputs for receiving the two input signals Si and s2, and with N outputs, N > 3, for providing a multiphase system of N baseband signals formed to be phase-shifted by T / N between them. In the case of sinusoidal optical signals, the period T of which can be defined angularly at 2ir as indicated previously, and for a multimodal coupler with three outputs such as that illustrated as a simple non-limiting example in [Fig.l], a three-phase system of three sinusoidal optical signals s'i, s'2 and s'3 phase-shifted by 2ir / 3 between them is provided.

[0030] The phase shift estimation device 22 further comprises, in the case of the provision of a three-phase optical system, an optical sensor associated with the multimodal coupler such as a camera provided with photoreceptors, for example a camera taking the form of a CCD (Charge Coupled Device) photographic sensor 26, for the provision of a three-phase electrical system S of three sinusoidal signals pb p2 and p3 phase-shifted by 2ir / 3. These three signals take the following form: p / t) ^a^osfOQ) ] +Oi ; p9(t) =a2cos[0(t)+2n / 3]+o2 ; p3(t)-a3cos[0(t)-2tt / 3] + o3.

[0031] In this three-phase system S, ab a2 and a3 are values of constant but a priori different amplitudes, which illustrates the fact that the three signals pb p2 and p3 are of different amplitudes because of imperfections in the laser source 14, the interferometer 12, the multimodal coupler 24 and / or the CCD sensor 26. Similarly, ob o2 and o3 are values of constant but a priori different amplitude shifts, which illustrates the fact that the three signals pb p2 and p3 are of different amplitude shifts because of imperfections in the laser source 14, the interferometer 12, the multimodal coupler 24 and / or the CCD sensor 26. As for 0(t), this is the phase shift time signal to be estimated which is coded as if it were the common frequency of the three signals pb p2 and p3 of the system three-phase S.

[0032] The three-phase system S is provided as input to a computer system 28 of the phase shift estimation device 22, this computer system 28 being specifically configured for processing the three-phase system S in order to extract therefrom an estimate of the phase shift between the two input signals Si and s2, as well as its evolution over time, in the form of a time signal SG of the sensorgram type, for example in the context of fluid characterization.

[0033] The computer system 28 as shown schematically in [Fig.l], comprises a processing unit 30 (i.e. a processor) conventionally associated with a memory 32 (for example a RAM memory or other).

[0034] This system can for example be implemented in a computer device such as a conventional computer comprising a processor associated with one or more memories for storing data files and computer programs whose instructions are intended to be executed by the processor. Their functions could also be at least partly microprogrammed or microwired in dedicated integrated circuits. Thus, as a variant, the computer device implementing the computer system 28 could be replaced by an electronic device composed solely of digital circuits (without a computer program) for carrying out the same actions.

[0035] This computer system 28 can for example carry out a two-stage processing, i.e. calibration then estimation of the phase shift by Clarke transformation or equivalent recalibrated, as taught in the aforementioned Halir et al document or in the document WO 2022 / 238170 A1 also cited above. This processing is defined and configured in the form for example of instructions of a software module or computer program 34 recorded in memory 32 and executable by the processing unit 30. More precisely, the calibration carried out by the aforementioned processing is defined to extract directly or indirectly, from the three-phase system S itself, calibration parameters as mentioned previously and to record them in memory 32. In other words, this is a blind calibration.

[0036] According to the teaching of the document Halir et al, the processing unit 30 can be configured using the computer program 34 to: - carry out a Clarke, Concordia or Fortescue transformation of the three-phase system S and deduce parameters of an ellipse fitted to this Clarke, Concordia or Fortescue transformation, - carry out the calibration of this adjusted ellipse by parametric transformation into a recentered circle, in which the calibration parameters include the parameters of the transformation, then - estimate the phase shift between the two input signals after applying the calibration parameters to the Clarke transformation of the three-phase system S.

[0037] According to the teaching of document WO 2022 / 238170 A1, the processing unit 30 can be configured using the computer program 34 to: - directly carry out the calibration of the N periodic signals of the three-phase system S from amplitude values and amplitude shifts of these N periodic signals, in which the calibration parameters comprise a combination of these amplitude values and amplitude shifts; - perform a Clarke, Concordia or Fortescue transformation of the three-phase system S after calibration and deduce parameters of a centered circle adjusted on this Clarke, Concordia or Fortescue transformation, then - estimate the phase shift between the two input signals from this Clarke, Concordia or Fortescue transformation.

[0038] But as a variant and advantageously, the computer system 28 comprises means for filtering the three-phase system S to recursively estimate a state of this system from a measurement. This is therefore Kalman-type filtering. These filtering means can also be defined and configured in the form of instructions of the software module or computer program 34 recorded in memory 32 and executable by the processing unit 30. They are more precisely cleverly configured so that the state to be estimated comprises the constant values of amplitudes ab a2 and a3 and of amplitude shifts ob o2 and o3 of the three sinusoidal signals pb p2 and p3 of the three-phase system S, a variable 0 relating to the phase shift 0(t) to be estimated and a constant value ô0 to aid in the estimation and monitoring of the variable 0, while the measurement comprises the three sinusoidal signals of the three-phase system S as supplied by the multimodal coupler 24 and the CCD sensor 26.By optimizing the state model by comparison with the measurement according to an error minimization criterion, for example the minimization of a mean square error in a well-known way in recursive filtering, the blind calibration is carried out, by estimation of the constant parameters ai, a2 and a3, ob o2 and o3 and ô0, at the same time as the phase shift estimation, by time tracking of the variable 0. As a constant parameter to help with the estimation and tracking of 0, ô0 is intended to converge towards the constant value equal to 0. As indicated previously, this time tracking as modeled by filtering ensures a continuity which makes it possible to do without phase unfolding.

[0039] Experimentally, it is observed that the estimation of the phase shift 0 presents much better results by the proposed recursive filtering than in the state of the art where the calibration is carried out prior to the estimation and separately. The results are also all the better when the initialization of the parameters ab a2, a3, ob o2, o3, 0 and ô0 is relevant. This is the case when we already have values ab a2 and a3, ob o2, o3 resulting from a previous calibration / estimation and when we initialize 0 and ô0 to 0. By default, the parameters ab a2, a3, ob o2 and o3 can be initialized to any predefined values.

[0040] Concretely and by way of non-limiting example, the filtering means can be in the form of a filter with two transfer functions: - a transition function F calculating a predicted value from an already estimated value of the state of the multiphase system, and

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048] - an observation function H calculating an estimate of the measurement from the predicted or already estimated value of the state of the multiphase system. This filter is then advantageously defined in a general way so that: - the state of the multiphase system is presented in the form of a vector whose coefficients include the 2N constant values of amplitudes and amplitude shifts of the N sinusoidal signals of the multiphase system, the variable value relating to the phase shift to be estimated and the constant value to aid estimation of the variable value, i.e. for example a vector with 2N+2 coefficients, and - the measurement is presented in the form of a vector with N coefficients, the N coefficients of which respectively comprise the N sinusoidal signals of the multiphase system as supplied by the multimodal coupler 24 and the CCD sensor 26. To simplify the notations, in the N = 3 case of the time-sampled three-phase system S, the state vector at a sampling time k can be written in the form: 1 H] 1 HE 01 °2 .oJj, The transition function F therefore makes it possible to model the link between Xk and Xk+i, for example according to the general equation: Xk+1 = F(Xk, Vk ), where Vk is a vector modeling processing noise. More precisely, a simple model of the link between Xk and Xk+i can be expressed as follows: Xk+) = F(Xk ) + Vk, where Vk has the same dimensions as Xk and is a Gaussian white noise vector. Advantageously, in view of the definition given of the state vector in the present invention, the transition function is defined as follows:

[0049] F(Xk) = 0 । ki °3 ik । 0 । k, o3,

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] This results in the predicted value Xk+i of the state vector being equal to the sum of the already estimated value Xk of the state vector, in which the coefficient comprising the variable value relating to the phase shift to be estimated is updated by adding the value of the coefficient comprising the constant value to aid estimation of the variable value, and the processing noise vector Vk. The observation function H, calculating an estimate of the measurement from the predicted or already estimated value of the state of the three-phase system S, makes it possible to model the link between Xk or Xk+[ and this estimate Yk+i of the measurement, for example according to the following general equation in the case of an estimate of the measurement from the already estimated value of the state of the three-phase system S: Yk+1 = H(Xk,Nk), where Nk is a vector modeling measurement noise. More precisely, a simple model of the link between Xk and Yk+[ can be expressed as follows: Yk+1 = H(Xk) + Nk, where Nk has the same dimensions as Yk +[ and is a Gaussian white noise vector. Advantageously and according to a first embodiment in which the filtering means 34 comprise an unscented Kalman filter UKF (from the English “Unscented Kalman Filter”), the observation function is defined in the following way for the three-phase system S: ■p^Xk)' H(Xk)= P2(Xk) a^ostO] +Oi = a2cos[0 + 27r / 3] + o 2 . a3cos[0-2jT / 3] + o3 It follows that the estimate of the measurement Yk+[ is equal to the sum of a vector with N=3 coefficients, the three coefficients of which are respectively the three signals of the three-phase system S expressed as a function of the coefficients of the already estimated value Xk of the state vector, and of the measurement noise vector Nk. Advantageously also and according to a second embodiment in which the filtering means 34 comprise an extended Kalman filter EKF (from the English "Extended Kalman Filter"), the observation function can be defined as follows: following for the three-phase S system:

[0059] H(X k ) = âp / XO axk dp2(xk> ùxk apjxk) 3Xk , Or

[0060] axk ^4X0 ùxk dP3(Xk) ax,. , cos[0] 0 0 cos[0+2tt / 3] 0 0 0 -ajsinfe] 0 100 0 - a2 sin[0+2tt / 3] 0 0 10 -2tt / 3] -a3 sin[6-27r / 3] 0 00 1]^

[0061] It follows that the estimation of the measurement Yk+i is equal to the sum of the matrix product of a Jacobian matrix of the three signals of the three-phase system S expressed as a function of the coefficients of the already estimated value Xk of the state vector with the already estimated value Xk of the state vector, and of the measurement noise vector Nk.

[0062] In accordance with the general principles of recursive filtering, the estimation of the parameters of the state of the three-phase system S, i.e. the estimation of the coefficients of the state vector in the case of a vector and matrix representation of the filtering, and the monitoring of their possible evolution over time is done by comparing the measurement and its estimation, more precisely by minimizing an estimation error by adapting the parameters accordingly. As a result, the parameters allowing the automatic calibration of the measurements provided are estimated jointly with the phase shift parameter, hence the provision of a simple and effective means for merging the calibration and the phase shift estimation, the latter producing the time signal SG of sensorgram type for example in the context of fluid characterization.It will be noted that the aforementioned filtering also provides that the calibration parameters aB a2 and a3, ob o2, o3 of the three-phase system S, estimated at the same time as the phase shift 0, are recorded in memory 32.

[0063] In accordance with the general principles of the present invention, and regardless of the embodiment chosen for the calibration and estimation of the phase shift between the two input signals S1 and S2, in particular among the three aforementioned teachings, the processing unit 30 executing the computer program 34 is configured to test whether the N signals of a multiphase system meet predetermined conditions required to carry out the estimation of the calibration parameters and to: - carry out the estimation of the calibration parameters and their recording in memory 32 if the required predetermined conditions are met, and - extract, from memory 32, the calibration parameters of a previous estimate if the required predetermined conditions are not met.

[0064] For example, if the calibration is carried out according to the teaching of the document Halir et al, that is to say by recalibration on a centered circle of the ellipse obtained by Clarke transformation or equivalent of a multiphase system, the predetermined conditions may relate to the supposed possibility of carrying out a reliable elliptical adjustment making it possible to obtain such an ellipse. It is within the reach of those skilled in the art to design an implementation thereof. But according to an advantageous and clever embodiment, these required predetermined conditions may comprise a predefined minimum number of sufficiently different phase shift values produced by Clarke transformation or equivalent of the multiphase system, for example according to a predefined model of distance between them and / or partitioning, to define and obtain a single ellipse solution by elliptical adjustment.In other words, if this minimum number is reached, then the elliptical adjustment is considered possible, reliable and unique solution, so that the calibration by elliptical adjustment and recalibration on a centered circle, the recording in memory 32 of the calibration parameters thus obtained, then the phase shift estimation can be carried out on the basis of the multiphase system provided. On the other hand, if this minimum number is not reached, the calibration parameters of a previous calibration are extracted from the memory 32 and applied to the multiphase system provided to carry out the phase shift estimation by Arctan calculation and phase unwinding of the Clarke transformation of the multiphase system.

[0065] For example also, if the calibration is carried out according to the teaching of document WO 2022 / 238170 A1 or the recursive filtering detailed above, that is to say directly from the amplitude and amplitude shift values of a multiphase system, the predetermined conditions may relate to the assumed ability to find these values. It is within the reach of those skilled in the art to design an implementation thereof. But according to an advantageous and clever embodiment, these required predetermined conditions may include the fact that each signal of the multiphase system provided results from enough different phase shift values, for example according to a predefined model of distance between them and / or partitioning, so that they cover at least the period T, that is to say the period 2ir in the case of sinusoidal signals.It is then possible to reconstruct, at least by interpolation and for each signal of the multiphase system, a single sinusoidal signal passing as close as possible to this signal over at least one angular period covered, so that it becomes possible to deduce its extreme values. therefore the values of its amplitude and its amplitude offset. Alternatively, the predetermined conditions required may include the fact that each signal of the supplied multiphase system varies sufficiently rapidly in amplitude to deduce its achievable extreme values. In other words, this means that it includes a sinusoidal time portion over at least one angular period allowing, at least by sinusoidal interpolation also, to estimate its extreme values.Thus, if at least sinusoidal interpolation is possible, then the estimation of the extreme values from which the amplitude and the amplitude shift of each signal of the provided multiphase system can be estimated is considered possible, reliable and unique solution, such that the calibration, the recording in memory 32 of the calibration parameters thus obtained, then the phase shift estimation can be carried out on the basis of the provided multiphase system (either successively in the case of document WO 2022 / 238170 A1, or simultaneously in the case of the recursive filtering detailed above).On the other hand, if the estimation of the extreme values of the signals of the provided multiphase system is not possible, the calibration parameters of a previous calibration are extracted from the memory 32 and applied to the provided multiphase system to carry out the phase shift estimation: by Clarke transformation of the multiphase system, Arctan calculation and phase unfolding in the case of document WO 2022 / 238170 A1; by recursive filtering by imposing the amplitude and amplitude shift parameters of the previous calibration for the recursive filtering detailed above.

[0066] The interferometric system 40 shown schematically in [Fig. 2] comprises a shared plurality of phase shift estimation devices such as that of [Fig. 1]. The common elements taken from the interferometric system 10 bear the same reference.

[0067] Thus, this system 40 comprises a plurality of interferometers 12i... 12n which all receive the sinusoidal optical signal s0 generated and emitted by the laser source 14 and each provide two input signals, denoted respectively s^, Si>2... sn>i, sn>2, to one of a plurality of devices 22p.. 22n for estimating their respective phase shifts, with n > 2. Each interferometer 12;, l <i<n, comporte avantageusement un dispositif d’interaction qui lui est propre dans sa branche optique d’interaction de sorte que les n déphasages obtenus sont respectivement propres aux n interféromètres 12J... 12n.

[0068] The n phase shift estimation devices 22p.. 22n respectively comprise n multimodal couplers 24i... 24n with two inputs, for receiving the n pairs of input signals su, Si,2... sn,b sn>2, and with three outputs for respectively supplying n three-phase systems of three sinusoidal signals s'i.p s'1>2, s'b 3... s'n>1, s'n>2, s'n> 3.

[0069] In the case of the provision of optical three-phase systems, the n phase shift estimation devices 22P.. 22n share the same CCD photographic sensor 26 associated with the n multimodal couplers 24P.. 24n which provide it with the n three-phase systems of three sinusoidal signals s'i l, s'i_2, s'i, 3... s'n.i, s'n>2, s'n> 3, for the provision of a set S of n electrical three-phase systems in each of which the three sinusoidal signals are phase-shifted by 2ir / 3. These 3n signals take the following form: p u (t) ^a^cos^a)] +o M ; P i2 (t) =3^2008(0^1) +2tt / 3] +o î>2 ; p x 3 (t) = ajjCOsfOjd) -2tt / 3] +o t3 ; P^Ct) =a 1M cos[0 n (t) ] +o iU ; p n2 ( t ) = a^cos[ 0 n ( t ) + 2tt / 3 ] + o n>2 ; P^Ct) = a^costedt) -2tt / 3] + .

[0070] The n phase shift estimation devices 22p.. 22n finally share the same computer system 28 specifically configured for processing the set S of the n three-phase systems provided in order to extract therefrom an estimate of the n respective phase shifts between the pairs of input signals s^, Si,2... sn.i, sn>2, as well as their respective evolutions over time, in the form of a set of n time signals SG;, l <i<n, par exemple constitué de n sensorgrammes en contexte de caractérisation de fluide. Le fonctionnement du système informatique 28 de la [Fig.2] est inchangé par rapport à ce qui a été détaillé en référence à la [Fig.l], moyennant une simple adaptation pour traiter simultanément les n systèmes triphasés fournis au lieu d’un seul.This is the reason why it bears the same reference 28 and comprises the same processing unit 30 associated with the memory 32 in which the aforementioned processing means are recorded (in accordance with the teaching of the document Halir et al, of the document WO 2022 / 238170 A1 or of the recursive filtering detailed above) in the form of the software module or computer program 34 capable of executing this time on the set S of the n three-phase systems provided, as well as the estimated calibration parameters.

[0071] The operation of the interferometric system 10, or the equivalent of the interferometric system 40, will now be detailed in a general manner with reference to [Fig.3].

[0072] During a prior interferometry step 100, the sinusoidal optical signal s0 generated and emitted by the laser source 14 passes through the interferometer 12 or the plurality of interferometers 12i... 12n., to provide the two input signals sb s2 or the 2n input signals Si4, Si>2... sn4, sn>2.

[0073] During a first step 102 of the method for estimating a phase shift between the two input signals sb s2 or between the 2n input signals Su, Si,2... sn.i, sn>2 two by two, these input signals undergo a multimodal coupling, possibly combined with a CCD capture, to obtain the three-phase system S of three sinusoidal signals phase-shifted from each other by 2ir / 3 (or more generally a multi-phase system of N signals phase-shifted from each other by T / N where N > 3) or the set S of n three-phase systems in each of which the three sinusoidal signals are phase-shifted from each other by 2ir / 3 (or more generally n multi-phase systems in each of which the N signals are phase-shifted from each other by T / N where N > 3).

[0074] During a following test step 104, the processing unit 30 executes the computer program 34 to test whether the predetermined conditions required to carry out an estimation of the calibration parameters from the three-phase system S or the set S of n three-phase systems provided are met. These predetermined conditions have been detailed previously and will not be repeated here. It will simply be noted that step 104 may require processing of the three-phase system S or the set S of n three-phase systems such as a Clarke transformation or equivalent in the case where the teaching of the Halir et al document is followed to carry out the calibration and the phase estimation.

[0075] If the predetermined conditions are met at the end of this test 104, the phase shift estimation method proceeds: - at a calibration step 106, including the estimation of the calibration parameters, and estimation of phase shift(s) between the two input signals sb s2 or between the 2n input signals si>b sb2... sn>b sn>2 two by two, and - at a step 108 of recording in memory 32 the calibration parameters estimated at step 106.

[0076] Given that step 108 depends on the success of test step 104 and on the fact that step 106 actually proceeds to estimate the phase shift parameters(s) by direct or indirect processing of the three-phase system S or of the set S of n three-phase systems provided, a logic function “AND” symbolized by the reference “&” is interposed between steps 106 and 108 of [Fig.3] to illustrate the verification of these two conditions.

[0077] If the predetermined conditions are not met at the end of the test 104, the phase shift estimation method moves to a second test step 110 during which checks whether calibration parameters from a previous estimate are available in memory 32.

[0078] If this is the case, the method proceeds to a step 112 of extracting these calibration parameters from the memory 32 and the method proceeds to step 106 during which the estimation of the calibration parameters by direct or indirect processing of the three-phase system S or of the set S of n three-phase systems provided is replaced by an application of the calibration parameters extracted in step 112. Otherwise, the method proceeds directly to step 106 during which the estimation of the calibration parameters by direct or indirect processing of the three-phase system S or of the set S of n three-phase systems provided is replaced by an application of calibration parameters defined by default for operation in degraded mode, for example operation without calibration when the calibration parameters defined by default are neutral arithmetic values.Alternatively, if no calibration parameter is available in memory 32 during test step 106, the phase shift estimation method could be stopped.

[0079] More specifically, when the calibration of the three-phase system S or of the set S of n three-phase systems provided and the estimation of phase shift(s) are separated, that is to say in particular when the teaching of the document Halir et al or of the document WO 2022 / 238170 A1 is applied, the operation of [Fig.3] can be specified in accordance with the diagram of [Fig.4].

[0080] In this figure, steps 100, 102, 104, 106, 108, 110 and 112 of [Fig.3] remain unchanged, but step 106 of calibration and estimation of phase shift(s) can be subdivided into three sub-steps 106A, 106B and 106C.

[0081] Sub-step 106A is only executed if the predetermined conditions are met at the end of test step 104. According to the teaching of the document Halir et al, it may consist of deducing, from the Clarke transformation(s) or equivalent transformation(s) carried out on the three-phase system S or on the set S of n three-phase systems in test step 104, the parameters of one or more ellipse(s) adjusted on this or these transformation(s), then carrying out a calibration of this or these adjusted ellipse(s) by parametrized transformation(s) into recentered circle(s), in which the calibration parameters comprise the parameters of the transformation(s).Alternatively, according to the teaching of document WO 2022 / 238170 A1, it may consist of carrying out a direct calibration of the three periodic signals of the three-phase system, or of the 3n periodic signals of the set S of n three-phase systems, from the amplitude and amplitude shift values of these three or 3n periodic signals, in which the calibration parameters comprise a combination of these amplitude and amplitude shift values.

[0082] Sub-step 106B is executed following step 106A or following step 112. It consists of applying the calibration parameters, obtained respectively in sub-step 106A or in step 112, to the adjusted ellipse or ellipses (according to Halir et al), or to the three-phase system S or to the set S of n three-phase systems (according to WO 2022 / 238170 A1).

[0083] Sub-step 106C is executed following step 106B or following step 110. According to the teaching of the document Halir et al, it consists of estimating the phase shift(s) on the adjusted and calibrated ellipse(s), then carrying out the phase unfolding(s). According to the teaching of the document WO 2022 / 238170 A1, it consists of carrying out a Clarke transformation or equivalent of the three-phase system S or of the set S of n calibrated three-phase systems, estimating the phase shift(s) on the adjusted circle(s), then carrying out the phase unfolding(s).

[0084] More specifically also, when the calibration of the three-phase system S or of the set S of n three-phase systems provided and the estimation of phase shift(s) are simultaneous, that is to say in particular when the teaching of the recursive filtering detailed above is applied, the operation of [Fig.3] can be specified in accordance with the diagram of [Fig.5].

[0085] In this figure, steps 100, 102, 104, 106, 108, 110 and 112 remain unchanged, but step 106 of calibration and estimation of phase shift(s) is executed in accordance with a variant 106' or 106” depending on whether the required predetermined conditions are met or not at the end of the test step 104.

[0086] In accordance with the two variants, the three-phase system or set of n three-phase (or more generally multi-phase) systems S is filtered to extract an estimate of the phase shift between the two input signals sb s2 or between the 2n input signals Su, Si,2... sn.i, this filtering being more precisely of the type to estimate a state of this (or each) three-phase (or more generally multi-phase) system from a measurement, this filtering being more precisely defined by the fact that: - the state to be estimated includes constant values of amplitudes and amplitude shifts of the three signals of the (or each) three-phase system (or more generally of the N signals of the (or each) multi-phase system), a variable relating to the phase shift to be estimated and a constant value to aid estimation of the variable value; and - the measurement includes the three signals of the (or each) three-phase system (or more generally the N signals of the (or each) multi-phase system) as provided by the multimodal coupling 102.

[0087] According to variant 106', the calibration parameters are estimated at the same time as the estimation of phase shift(s) by recursive filtering. They are for example initialized at default values. But they can also be initialized to calibration parameter values from a previous estimate if any exist in memory 32.

[0088] According to variant 106”, the calibration parameters are imposed at default values, following step 110, or at calibration parameter values of a previous estimation, following step 112, which places a constraint on the recursive filtering carried out, the latter then no longer being authorized to converge the constant values of amplitudes and amplitude shifts of the three signals of the (or of each) three-phase system (or more generally of the N signals of the (or of each) multi-phase system) towards values other than those of the imposed calibration parameters.

[0089] Of course, step 108 can then only be carried out if it is variant 106' of step 106 which is executed.

[0090] The interferometric system 40 of [Fig. 2] can be implemented in an electronic device 50 for olfactory characterization of a fluid such as that shown schematically in [Fig. 6]. This is a non-limiting example of a fluid characterization device according to a possible embodiment of the present invention for a non-limiting olfactory application of odor identification. It comprises a measuring chamber 52 intended to receive a fluid, for example a gas such as ambient air. To do this, it comprises a suction device 54 designed to suck in the air located inside the measuring chamber 52 and cause it to exit to the outside. It further comprises an air inlet 56 which can be selectively closed to keep the ambient air in the measuring chamber 52 or open to allow the evacuation of the ambient air from the measuring chamber 52 and its renewal by activating the suction device 54.It is thus equipped with means of controlling incoming and outgoing flows.

[0091] In its measuring chamber 52, the electronic olfactory characterization device 50 comprises several sensors, in particular olfactory sensors 58, distributed respectively over as many reactive sites, for example around sixty, designed to interact with compounds likely to be present in the measuring chamber 52 when the device 50 is placed near a fluid to be analyzed emitting these compounds, in particular when the air inlet 56 is near the fluid considered. The compounds emitted are generally volatile organic compounds but the present invention is not limited to such compounds.

[0092] Each olfactory sensor 58 is itself, for example, a biosensor designed to interact with the compounds of a particular family of volatile organic compounds. In practice, each olfactory sensor 58 may comprise a molecule, such as a peptide immobilized on a substrate or a polymer covering a surface, complementary to the compounds of the family associated with this olfactory sensor 58. The idea The general approach is to functionalize reactive sites using olfactory sensors 58 (i.e. biosensors, polymers, carbon nanotubes, etc.) such that they differentially adsorb and desorb volatile organic compounds, distort a differentiated molecular interaction response of the olfactory sensors, and amplify the response in the form of a sequence S of electrical measurement signals using a physical transduction device.

[0093] Alternatively, the electronic olfactory characterization device 50 could be adapted to be brought into contact with any other fluid, liquid or gas, than ambient air. It could also, according to a particularly simple version, not include the suction device 54 and the air inlet 56, or even the measuring chamber 52. In this simple version, the olfactory sensors 58 are then capable of being directly brought into contact with the fluid to be analyzed without flow control.

[0094] The olfactory sensors 58 are associated with at least one transducer 60 with which they interact. This transducer 60 is arranged and configured to measure any change in physical property caused by an interaction of the olfactory sensors 18 with the fluid to be analyzed. It provides the electrical measurement signal S which is presented for example in the form of a sequence of electrical measurement signals and characterizes this fluid since this sequence is representative of the volatile organic compounds with which the olfactory sensors 58 can interact in the measurement chamber 52.

[0095] More precisely, the transducer 60 is a system for processing a periodic signal, i.e. the periodic signal s0 previously defined in the interferometric system 40, by interferometry. More precisely, it comprises the plurality of interferometers 12i... 12n which all receive the periodic signal s0 and each have one of the sensors 58 as an interaction device of their interaction branch. There are therefore as many interferometers 12i... 12n in the transducer 60 as there are sensors 58.

[0096] As a non-limiting example, the transducer 60 is, as mentioned previously, a system for amplifying optical index variation by Mach-Zehnder interferometry, for example according to a Mach-Zehnder interferometer matrix technology, called MZI technology (from the English “Mach-Zehnder Interferometer”), and even more precisely a multimodal interference MZI technology, called MZI / MMI technology (from the English “Multi Mode Interference”), since the transducer 60 further comprises the n multimodal couplers 24i... 24n with two inputs, for receiving the n pairs of input signals Si4, Si>2... sn4, sn>2, and with three outputs for the respective supply of n three-phase systems of three sinusoidal signals s's'i>2, s'i>3... s'n.i, s'n>2, s'n>3, as well as the photographic sensor CCD 26 pooled to provide the set S of the n three-phase systems as an electrical measurement signal. This transducer 60 is therefore configured to measure any change in a refractive index due to an interaction of the fluid studied with at least any one of the olfactory sensors 58 thanks to the detectable phase shift between the reference arm and the detection arm of the interferometer on which this any olfactory sensor is arranged.

[0097] The electronic olfactory characterization device 50 further comprises several functional modules which will be described below. In the example described, these modules are of a software nature. Thus, the device 50 comprises the computer system 28 previously described, the latter comprising the processing unit 30 and the associated memory 32 in which several computer programs or several functions of the same computer program are recorded, including in particular the software module or computer program 34. These computer programs comprise instructions designed to be executed by the processing unit 30 in order to carry out the functions of the software modules. They are presented as distinct, but this distinction is purely functional. They could just as well be grouped according to all possible combinations into one or more software programs.Their functions could also be at least partly micro-programmed or micro-wired in dedicated integrated circuits, such as digital circuits. Thus, alternatively, the computer system 28 could be replaced by an electronic device composed solely of digital circuits (without a computer program) for carrying out the same functions.

[0098] The electronic olfactory characterization device 50 thus firstly comprises a software module 62, intended to be executed by the processing unit 30, for controlling the suction device 54 (if it is provided in the device 50), the air inlet 56 (if it is also provided in the device 50) and the transducer 60.

[0099] It further comprises, optionally but advantageously, a software module 64, intended to be executed by the processing unit 30, for selecting, from among the olfactory sensors 58 of the electronic olfactory characterization device 50, a subset of sensors sensitive to volatile components characteristic of a desired olfactory imprint. These characteristic volatile components may vary from one application or fluid studied to another so that the selection of olfactory sensors carried out by the software module 64 may also vary and be parameterized. The selected subset comprises, for example, M > 1 olfactory sensor(s), in particular advantageously several olfactory sensors (M > 2).

[0100] The electronic olfactory characterization device 50 further comprises the software module or computer program 34, already defined and described in detail, intended to be executed by the processing unit 30, to extract M sensorgrams SG;, i G {1, ..., M} respectively representative of the interactions of the M selected olfactory sensors with the volatile organic compounds concerned from the values specific to these M olfactory sensors in the sequence S of three-phase systems provided by the transducer 60.

[0101] Figure 7A thus illustrates the superimposed time diagrams of around sixty sensorgrams SG,, i G {1, ..., M} obtained over a period of approximately 190 seconds according to a well-controlled measurement protocol, involving control of the suction device 54 and the air inlet 56, in which: - the olfactory sensors 58 are first exposed to a reference fluidic environment with a carrier fluid without the presence of the target compounds of a fluid to be analyzed during a first reference state identifiable by a first PHI portion of the sensorgrams, - they are then exposed to the fluid to be analyzed during a second analytical adsorption state triggered by a controlled injection of this fluid into the measuring chamber 52, this second state being identifiable by a second portion PH2 of the sensorgrams, and - they are finally exposed again to the reference fluidic environment during a third final state of desorption by a controlled evacuation of the fluid to be analyzed from the measuring chamber 52, this third state being identifiable by a third portion PH3 of the sensorgrams.

[0102] Returning to Figure 6, the electronic olfactory characterization device 50 further comprises an optional software module 66, intended to be executed by the processing unit 30, to carry out possible prior processing on the M sensorgrams SGj, ig [ 1, ..., M} provided by the software module 34.

[0103] This pre-processing comprises for example a low-pass filtering implemented in the form of a digital filter with finite or infinite impulse response. This involves filtering the high-frequency measurement noise in the raw signals as provided by the software module 34. A Butterworth filter with a first-order finite impulse response and a cut-off frequency normalized to 0.45 (i.e. the value of the ratio between the cut-off frequency and the sampling frequency equal to 0.45) is suitable.

[0104] This preliminary processing further comprises, for example, a calculation of a norm within the meaning of patent document WO 2020 / 141281 A1 on the M sensorgrams SGj. i G {1, ..., M] filtered or not to obtain M filtered and / or normalized sensorgrams SG-, iG {1, ..., M)-

[0105] The electronic olfactory characterization device 50 further comprises a software module 68, intended to be executed by the processing unit 30, to obtain in a well-known and non-detailed manner a characterization or SIG signature of the composition of the fluid to be analyzed from the M SG sensorgrams? ie {1 ..., M] or SG-, ie {1.....M} • This characterization or SIG signature can take the form of a standardized olfactory signature as illustrated in [Fig.7B] in the form of a circular diagram. It should be noted that this module can proceed in two stages: firstly obtaining a first intermediate signature, then transforming this first intermediate signature by normalization.

[0106] The SIG signature illustrated in [Fig.7B] comprises sixty-four components. The software module 68 can therefore optionally but advantageously be further programmed to transform the SIG signature, whether standardized or not, into another simplified SIG' signature by component reduction. A method using linear discriminant analysis LDA (from the English "Linear Discriminant Analysis"), principal component analysis PCA (from the English "Principal Component Analysis"), independent component analysis ICA (from the English "Independent Component Analysis"), auto-encoder, etc., is suitable. For a SIG signature with sixty-four components such as that of [Fig.7B], a simplified SIG' signature with two or three components can be obtained.

[0107] A method of olfactory characterization of a fluid corresponding to the execution of software modules 34 and 62 to 68 will now be detailed in accordance with the succession of steps in [Fig.8].

[0108] During a first optional step 200 carried out by executing the software module 64, a part of the olfactory sensors 58 is selected according to a desired olfactory imprint.

[0109] During a following step 202, the electronic olfactory characterization device 50 is brought close to a fluid so as to obtain an interaction of the fluid with each of the selected olfactory sensors 58. On this occasion, the software module 62 is executed.

[0110] During a following step 204, the transducer 60 provides a sequence S of measurement signals in interaction with the olfactory sensors 58.

[0111] During a following step 206 carried out by executing the software module 34, the processing unit 30 provides the M sensorgrams SG? ie {1, ..., M] representative of an interaction of the fluid with each olfactory sensor 58 selected after having carried out a blind calibration and an estimation of phase shifts.

[0112] During a following optional step 208 carried out by executing the software module 66, the processing unit 30 provides the M sensorgrams SG'j, ig {1, ..., M].

[0113] Finally, during a last step 210 carried out by executing the software module 68, the processing unit 30 provides the signature SIG or SIG' from the M sensorgrams SGb iG {1, ..., M] or SG'^ i G {1, ..., M}.

[0114] It is clear that a phase shift estimation device such as one of those described previously with reference to FIGS. 1 and 2 makes it possible to improve blind calibration by providing a test step on conditions required to carry out the estimation of the calibration parameters as well as a fallback solution to the calibration parameters of a previous estimation if the test fails. This results in a better estimation of the phase shift. Such a device also finds a particularly advantageous application in the characterization of a fluid by the use of an electronic device such as that illustrated in [Fig.6].

[0115] As an illustration and in order to demonstrate the interest of such an improved calibration in the phase shift estimation device, an experiment was carried out on a three-phase system resulting from interferometry having a phase shift amplitude of approximately 2 radians, which is far from covering the required period 2ir. The three-phase system S as provided is shown at the top left of [Fig.9]. It has 3 signals pb p2, Pi depending on time and having variations in amplitude A (expressed without unit, in gray levels for example). By calibrating this three-phase system S using calibration parameters from a previous estimation, for example by following steps 110, 112, 106B and 106C of the method of [Fig.4], it is possible to obtain the calibrated three-phase system S' in which the signals p'b p'2 , p'3 are corrected in amplitudes and amplitude shifts.Although this calibration is not ideal, because it is not carried out on the three-phase system S supplied itself, it remains advantageous to provide for it in the event of failure of test 104 because we can clearly see that the sensorgram SG' obtained from S' is closer to the reality of the experiment than the sensorgram SG obtained from S. Indeed, S presents phase shift variations between 0 and 0.7 radians, which does not correspond to the experiment conducted, whereas S' evolves well between 0 and 2 radians.

[0116] It will also be noted that the invention is not limited to the embodiments described above. It will indeed appear to those skilled in the art that various modifications can be made to the embodiments described above, in light of the teaching which has just been disclosed to them. In the detailed presentation of the invention which is given above, the terms used should not be interpreted as limiting the invention to the embodiments set out in the present description, but should be interpreted to include all equivalents whose prediction is within the reach of those skilled in the art by applying their general knowledge to the implementation of the teaching which has just been disclosed to them.

Claims

Claims

1. Device (22; 22b.. 22n) for estimating a phase shift between two input signals (sb s2; Si >b Si>2... sn>b sn>2) from the same periodic signal (s0) of period T processed by interferometry, comprising: - a multimodal coupler (24; 24i... 24n) with two inputs, for receiving the two input signals (sb s2; Si j, sb2... sn>b sn 2), and with N outputs, N > 3, for providing a multiphase system (S) of N signals formed to be phase shifted by T / N between them; - a unit (30) for processing the multiphase system (S) to extract calibration parameters and an estimate (SG; SG;) of the phase shift between the two input signals (si, s2; Si, b sb2...sn,b sn>2); and - means (32) for storing the calibration parameters; characterized in that the processing unit (30) is configured (34) to test whether the N signals of the multiphase system (S) meet predetermined conditions required to carry out an estimation of the calibration parameters and to: - carry out the estimation of the calibration parameters and record them in the storage means (32) during the execution of the processing of the multiphase system (S) if the predetermined required conditions are met; and - extract, from the storage means (32), the calibration parameters of a previous estimation and reuse them during the execution of the processing of the multiphase system (S) if the predetermined required conditions are not met.

2. Device (22; 22b.. 22n) for estimating a phase shift according to claim 1, in which the predetermined required conditions that the processing unit (30) is configured (34) to test comprise the fact that each signal of the multiphase system (S) provided results from enough different phase shift values so that they cover at least the period T.

3. Device (22; 22b.. 22n) for estimating a phase shift according to claim 2, in which the test carried out by the processing unit (30) comprises a verification that each signal of the system multiphase (S) supplied varies sufficiently in amplitude to deduce its extreme values which can be reached.

4. A phase shift estimation device (22; 22p.. 22n) according to any one of claims 1 to 3, wherein the predetermined required conditions that the processing unit (30) is configured (34) to test include that a Clarke, Concordia or Fortescue transformation of the provided multi-phase system (S) produces enough different phase shift values to define a single elliptic fit ellipse solution.

5. Device (22; 22p.. 22n) for estimating a phase shift according to any one of claims 1 to 4, in which the processing unit (30) is provided with means (34) for filtering the multiphase system (S) designed to estimate a state of this multiphase system (S) from a measurement, these filtering means (34) being more precisely defined by the fact that: - the state to be estimated comprises constant values of amplitudes and amplitude shifts of the N signals of the multiphase system (S), a variable value relating to the phase shift to be estimated and a constant value for assisting in estimating the variable value; and - the measurement comprises the N signals of the multiphase system (S) as supplied by the multimodal coupler (24; 24i...24n); and in which: - the calibration parameters comprise the constant values of amplitudes and amplitude shifts of the N signals of the multiphase system (S); - the processing unit (30) is designed to extract the estimate of the phase shift on the basis of the variable value relating to the phase shift of the state to be estimated by filtering the multiphase system (S).

6. Device (22; 22p.. 22n) for estimating a phase shift according to any one of claims 1 to 4, in which the processing unit (30) is configured to: - carry out a Clarke, Concordia or Fortescue transformation of the multiphase system (S) and deduce therefrom

7.

8. parameters of an ellipse fitted to this Clarke, Concordia or Fortescue transformation; - carry out a calibration of this adjusted ellipse by parametric transformation into a recentered circle, in which the calibration parameters include the parameters of the transformation; then - estimate the phase shift between the two input signals (si , s2 ; Si i, Si,2... sn.i, sn>2) after applying the calibration parameters to the Clarke transformation of the multiphase system (S). Device (22; 22p.. 22n) for estimating a phase shift according to any one of claims 1 to 4, in which the processing unit (30) is configured to: - carrying out a calibration of the N periodic signals of the multiphase system (S) from amplitude values and amplitude shifts of these N periodic signals, in which the calibration parameters comprise a combination of these amplitude values and amplitude shifts; - perform a Clarke, Concordia or Fortescue transformation of the multiphase system (S) after calibration and deduce parameters of a centered circle adjusted on this Clarke, Concordia or Fortescue transformation; then - estimate the phase shift between the two input signals (sb s2 ; Sj i, s1>2... sn4, sn>2) from this Clarke, Concordia or Fortescue transformation. Electronic device (50) for characterizing a fluid, comprising: - at least one sensor (58) designed to interact with the fluid; and - a transducer (60) designed to provide, in interaction with said at least one sensor (58), a sequence of at least one measurement signal (S) representative of an interaction of the fluid with each sensor (58); in which the transducer (60) is a system for processing a periodic signal (s0) by interferometry and the electronic device characterization device (50) comprises at least one device (22; 22i... 22n) for estimating a phase shift between two input signals (sb s2; Si,i, s1>2... sn,b sn>2) from said periodic signal (s0) according to any one of claims 1 to 7.

9. Electronic device (50) for characterizing a fluid according to claim 8, in which the transducer (60) is a system for amplifying optical index variation by Mach-Zehnder interferometry with multimodal interference, the multimodal coupler (24; 24i • • • 24n) then having optical outputs and associated with an optical sensor (26).

10. Method for estimating a phase shift between two input signals (sb s2; Si i, sb2... sn.b sn>2) from the same periodic signal (s0) of period T processed by interferometry, comprising: - a multimodal coupling (102) of the two input signals (sb s2; Si i, sb2... sn.b sn>2) for providing a multiphase system (S) of N signals formed to be phase-shifted from each other by T / N where N > 3; - a processing (104, 106, 110, 112) of the multiphase system (S) to extract calibration parameters and an estimation (SG; SGi) of the phase shift between the two input signals (sb s2; Sj >b Sj,2... sn>1, sn>2); and - a recording (108) of the calibration parameters in storage means (32); characterized in that the processing (104, 106, 110, 112) comprises a test (104) on predetermined conditions required to perform an estimation of the calibration parameters and is adapted to selectively perform one or other of the following operations: - the estimation (106; 106A; 106') of the calibration parameters and their recording (108) in the storage means (32) during the execution of the processing of the multiphase system (S) if the predetermined conditions required are fulfilled by the N signals of the multiphase system (S); and - extracting (112), storage means (32), calibration parameters from a previous estimate and reusing them when executing the system processing multiphase (S) if the required predetermined conditions are not met by the N signals of the multiphase system (S).

Images