METHOD FOR MONITORING A CYLINDER REFRIGERATOR BY PROGRESSIVE MECHANICAL SHAFT MEASUREMENT

Description

DOMAINE TECHNIQUE

[0001] The invention relates to the field of monitoring a mechanical transmission device in an aircraft. ÉTAT DE LA TECHNIQUE ANTÉRIEURE

[0002] Various systems and methods for monitoring a mechanical transmission device in an aircraft are known in the prior art.

[0003] Patent application EP3217170 teaches monitoring such a mechanical transmission device using an acoustic sensor. The acoustic sensor measures an acoustic signal generated by mechanical vibrations in the mechanical transmission device. An anomaly in the operation of the mechanical transmission device results in the appearance of characteristic peaks in the frequency spectrum of the measured acoustic signal.

[0004] An objective of the present invention is to provide an improved method for monitoring a mechanical transmission device in an aircraft, which is particularly suitable for monitoring an epicyclic gear train.

[0005] We can also cite patent application CN 105 806 613 A and patent US 10 168 248 B1. EXPOSÉ DE L'INVENTION

[0006] This objective is achieved with a method for monitoring an epicyclic gear train of an aircraft, in which the epicyclic gear train comprises two first cogwheels mounted coaxially and at least one second cogwheel meshing with the first two cogwheels, the method comprising the following steps: acquisition, at a predetermined sampling frequency, of first values ​​of a signal formed by a progressive mechanical wave generated in the epicyclic gear train, said first values ​​forming primary mechanical wave data; measurement, at a plurality of successive instants, of values ​​of a rotation speed of at least one of the two first toothed wheels and the at least one second toothed wheel, said measured values ​​forming primary rotation data; from the primary rotation data, calculation of values ​​of a rotation speed of a point of contact between the one first toothed wheel and the one second toothed wheel of the epicyclic gear train, said calculated values ​​forming secondary rotation data;from the primary mechanical wave data and the secondary rotation data, determining second values ​​of said signal formed by a progressive mechanical wave generated in the epicyclic gear train, said second values ​​corresponding to regular sampling as a function of the phase of said contact point and forming secondary mechanical wave data; and using the secondary mechanical wave data to detect an anomaly related to the operation of the epicyclic gear train.;

[0007] The first two cogwheels mounted coaxially are called respectively the crown and the sun, where the sun is surrounded by the crown. The at least one second cogwheel is called the satellite.

[0008] The primary mechanical wave data corresponds to a sampling, as a function of time, of a signal formed by a progressive mechanical wave generated in the epicyclic gear train. Said signal relates to a vibration in an elastic medium. It can be an acoustic signal, or a vibrational signal capable of being measured by an accelerometer. A time interval between two successive sampling instants is constant.

[0009] The primary rotation data are measured at different times, in parallel with the acquisition of the primary mechanical wave data. The primary rotation data designates measurements of the respective rotational speed of at least one of the toothed wheels of the epicyclic gear train. The rotational speed, or engine speed, is expressed for example in revolution(s) per unit of time. The rotational speed is measured for example using an optical sensor detecting the passage of a predetermined reference point on the toothed wheel.

[0010] Secondary rotation data are calculated values ​​of a rotational speed of a contact point between two toothed wheels in the epicyclic gear train. The rotational speed of said contact point forms a so-called fictitious regime. The contact point is a contact point between a satellite and the crown, or between a satellite and the sun. Any one of these contact points can be considered. The values ​​of a rotational speed of the contact point are determined for a plurality of instants, preferably corresponding to the instants of measurement of the primary rotation data.

[0011] The calculated values ​​of a rotation speed of the contact point are then used to simulate a resampling of the signal formed by the progressive mechanical wave generated in the epicyclic gear train. The resampling is this time constant depending on the phase of said contact point. In other words, there is a constant phase difference between two points of the resampling, where the phase relates to said contact point and is obtained from the rotation speed of said contact point. This new sampling of the signal formed by the progressive mechanical wave generated in the epicyclic gear train forms secondary mechanical wave data.

[0012] The secondary mechanical wave data are then used to detect an anomaly in the operation of the epicyclic gear train. For this purpose, models can be used, associating an anomaly with predefined characteristics of a function that depends on the secondary mechanical wave data. The models can be determined experimentally or by simulations.

[0013] The invention thus provides a method for monitoring an epicyclic gear train of an aircraft, based on clever resampling of a signal formed by a progressive mechanical wave generated in the epicyclic gear train.

[0014] It may be advantageous to resample in phase a signal formed by a progressive mechanical wave generated in a mechanical transmission device, when the latter is a simple gear with only two toothed wheels. In the case of an epicyclic gear train, the complex arrangement of the toothed wheels encourages the skilled person to avoid such resampling. The idea behind the invention consists of maintaining in-phase resampling, but taking as reference not one of the toothed wheels of the epicyclic gear train, but a point of contact between two of these toothed wheels. The inventors were able to show that this idea makes it possible to obtain resampled data allowing effective identification of a mechanical anomaly of the epicyclic gear train.

[0015] Advantageously, the rotational speed of the point of contact between a first toothed wheel and a second toothed wheel of the epicyclic gear train is a linear combination of the respective rotational speeds of said first and second toothed wheels.

[0016] At least one of said respective rotational speeds can be calculated from the primary rotational data ( V mes _ r ( t j )) and respective numbers of teeth of at least two of the first two gears and the at least one second gear.

[0017] According to an advantageous embodiment, the use of secondary mechanical wave data ( S ( ϕ k )) to detect an anomaly related to the operation of the epicyclic gear train includes: a calculation of analysis data, including a calculation of a Fourier transform of the secondary mechanical wave data ( S ( ϕ k )), where the Fourier transform transforms a function of a phase into a function of an order; and the use of said analysis data to detect an anomaly linked to the operation of the epicyclic train.

[0018] Analysis data associated with at least one whole order can then be used to detect an anomaly linked to the operation of the epicyclic gear train.

[0019] The calculation of analysis data may include the calculation of a power spectral density of the secondary mechanical wave data ( S ( ϕ k )).

[0020] Alternatively, the analysis data calculation may include calculating an averaged power spectral density of the secondary mechanical wave data ( S ( ϕ k )).

[0021] Alternatively, calculating analysis data may comprise calculating a spectrogram aggregating a plurality of values ​​of an averaged power spectral density of the secondary mechanical wave data ( S ( ϕ k )).

[0022] The signal formed by a progressive mechanical wave generated in the epicyclic gear train may be an acoustic signal, acquired using an acoustic wave sensor. This acoustic wave sensor is advantageously arranged in a position spaced from the epicyclic gear train.

[0023] Alternatively, the signal formed by a progressive mechanical wave generated in the epicyclic gear train is a vibration signal, acquired using an accelerometer.

[0024] The monitoring method may further comprise a maintenance step on the epicyclic gear train, when an anomaly linked to the operation of the latter is detected using the secondary mechanical wave data (S ( ϕ k )).

[0025] Finally, said monitoring method may further comprise a preliminary step of determining health indicators, which comprises the following sub-steps: for a plurality of flights, acquiring primary mechanical wave data and primary rotation data, and determining corresponding secondary rotation data and secondary mechanical wave data; for each of the plurality of flights, calculating analysis data, said calculation comprising calculating a Fourier transform of the secondary mechanical wave data; and correlating the analysis data with information relating to an anomaly or an absence of anomaly in the operation of the epicyclic gear train.

[0026] The invention also relates to a monitoring system for monitoring an epicyclic gear train of an aircraft, the system comprising: an epicyclic gear train which comprises two first cogwheels mounted coaxially and at least one second cogwheel meshing with the two first cogwheels; at least one mechanical wave sensor, arranged to acquire, at a predetermined sampling frequency, first values ​​of a signal formed by a progressive mechanical wave generated in the epicyclic gear train, said first values ​​forming primary mechanical wave data ( S ( t i )) ; at least one rotation speed sensor, arranged to acquire, at a plurality of successive instants, values ​​of a rotation speed of at least one of the two first toothed wheels and the at least one second toothed wheel, said measured values ​​forming primary rotation data ( V nes_r ( t j )); and a processor, configured to receive as input the primary mechanical wave data ( S ( t i )) and the primary rotation data ( V mes-r ( t j )), to calculate values ​​( V c ( t j )) of a rotation speed of a point of contact between a first toothed wheel and a second toothed wheel of the epicyclic gear train, to determine second values ​​of said signal formed by a progressive mechanical wave generated in the epicyclic gear train, said second values ​​being sampled as a function of a phase of said point of contact, and to provide as output information ( D ) relating to the presence or absence of an operating anomaly of the epicyclic gear train.

[0027] The mechanical wave sensor is typically an acoustic sensor, arranged at a position spaced from the epicyclic gear train. BRÈVE DESCRIPTION DES DESSINS

[0028] The present invention will be better understood by reading the description of exemplary embodiments given purely for informational purposes and in no way limiting, with reference to the appended drawings in which: THE figures 1 et 2 schematically illustrate an epicyclic gear train, in a perspective view, respectively a front view; the figure 3 schematically illustrates a method according to the invention, for monitoring the epicyclic gear train of the figure 1 ; THE figures 4A à 4E illustrate a resampling according to the invention; the figures 5A et 5B respectively illustrate the signal formed by a progressive mechanical wave generated in the epicyclic train of the figure 1 , as a function of time, and a corresponding power spectral density; the figures 6A et 6B respectively illustrate the same signal as in the figure 5A represented this time as a function of the phase of a contact point in the epicyclic gear train, and a corresponding power spectral density; the figures 7 et 8 illustrate detailed views of a spectrogram obtained using a method according to the invention, centered respectively on order 1 and on order 2; the figure 9 schematically illustrates a system according to the invention; and the figure 10 schematically illustrates a turbomachine of an aircraft receiving an epicyclic gear train as monitored in the invention. EXPOSÉ DÉTAILLÉ DE MODES DE RÉALISATION PARTICULIERS

[0029] There figure 1 illustrates, in a perspective view, an example of an epicyclic gear train 100 to which the invention can be applied. The epicyclic gear train 100 here comprises: a crown 110, mounted integrally with a downstream propeller of a turboshaft engine (not shown); a sun 120, mounted integrally with a shaft connected to a power turbine (not shown); and three satellites 130, mounted integrally with an upstream propeller of the turboshaft engine (not shown) by means of a three-branch connecting piece 131.

[0030] The crown 110 and the sun 120 are mounted coaxially, with the crown 110 arranged around the sun 120. Together they form two first toothed wheels.

[0031] The three satellites 130 are equally distributed around the sun 120, and each arranged between the ring gear 110 and the sun 130. Together they form three second toothed wheels which each mesh with both the ring gear 110 and the sun 130. The connecting piece 131 has each of its three branches fixed to the rotation shaft of one of the satellites 130.

[0032] There figure 2 illustrates, schematically, the epicyclic gear train 100 according to a front view. The direction of rotation of each of the elements of the epicyclic gear train 100 is represented by arrows. The respective directions of rotation of the crown 110 and the sun 120 are opposite. Each of the three satellites 130 can rotate on itself and / or around the sun 120. The rotation of the satellites 130 around the sun 120 is symbolized by figure 2 by a dotted circle. The direction of rotation of the satellites 130 around the sun 120 corresponds to the direction of rotation of the sun 120 on itself.

[0033] There figure 2 also illustrates two points of contact P C1 , respectively P C2 , in the epicyclic gear train 100. P C1 is the point of contact between one of the satellites 130 and the crown 110. P C2 is the point of contact between one of the satellites 130 and the sun 120. Each of the three satellites has two points of contact, respectively with the crown 110 and with the sun 120.

[0034] There figure 3 schematically illustrates a method according to the invention, for monitoring the epicyclic gear train 100 in operation.

[0035] In a first step 301, a series of values ​​is acquired S ( t i ) of a signal S(t) formed by a progressive mechanical wave generated in the epicyclic gear train 100. The progressive mechanical wave is in particular an elastic wave. It is generated in particular by the collisions of the teeth of the satellites with the teeth of the crown and the teeth of the sun. The characteristics of these collisions are a function in particular of manufacturing errors and the elastic deformation of the teeth. The signal S(t) is preferably an acoustic signal, relating to a wave radiated by the epicyclic gear train, and acquired using an acoustic wave sensor such as a microphone or an acoustic pressure sensor. Since an acoustic signal can be propagated by any solid, liquid or gaseous medium, the acoustic wave sensor can be freely arranged, directly on the epicyclic gear train or at a distance from the latter. Alternatively, the signal S(t) is a vibrational signal, relating to a wave conducted in an elastic solid medium, and acquired using a sensor such as an accelerometer. In this case, the accelerometer must be arranged in direct physical contact with the epicyclic gear train or with an interposed solid medium itself in direct physical contact with the epicyclic gear train. The values S ( t i ) are acquired with a sampling frequency f ech . In other words, the moments of acquisition t i associated with different values S ( t i ) are separated two by two by a fixed time interval. The acquired values S ( t i ) form data called primary mechanical wave data.

[0036] In a second step 302, implemented in parallel with step 301, a series of values ​​is acquired V mes_r ( t j ) of the respective rotation speed(s) of at least one of the toothed wheels r among the crown, the sun, and the three satellites. In other words, we acquire a series of values ​​of the respective speed of at least one of the toothed wheels r among the crown, the sun, and the three satellites of the epicyclic gear train. The rotation speed values V mes _ r ( t j ) are angular speeds, expressed for example in revolutions per minute. Each rotation speed here designates the rotation speed of a toothed wheel, on itself. In the case of a satellite 130, this rotation speed relates only to the rotation of the satellite 130 on itself, without consideration of its possible rotation around the sun 120. It is preferably assumed that the three satellites have the same rotation speed. The values V mes_r ( t j ) are acquired using a tachometer-type sensor, comprising for example an optical sensor capable of detecting the passage of a predetermined reference point on the toothed wheel, or a sensor capable of measuring the duration of passage of a toothed wheel (a short duration corresponding to a rapid passage of teeth, and therefore to a high rotation speed). Where appropriate, the rotation speed of a toothed wheel is measured at the level of a propeller or a turbine mounted integrally with the latter. The different values V mes _ r ( t j ) are associated with a series of acquisition times t j . The values V mes_rr ( t j ) form data called primary rotation data.

[0037] In a third step 303, the primary rotation data is used to calculate values V c ( t j ) of a theoretical rotational speed of a contact point in the epicyclic gear train 100. This rotational speed is also called a "fictitious regime". The contact point designates any one of the contact points between one of the satellites 130 and the crown 110, respectively between one of the satellites 130 and the sun 120. The figure 2 illustrates two of these contact points, P C1 and P C2 .

[0038] The rotational speed of the contact point is a rotational speed considered in a reference frame attached to an aircraft receiving the epicyclic gear train. The rotational speed is the same for all the contact points of the epicyclic gear train 100, so that one or the other of these contact points can be considered indifferently.

[0039] The rotational speed of the contact point between two toothed wheels of the epicyclic gear train is a linear combination of the respective rotational speeds of said two toothed wheels. It also depends on the respective numbers of teeth of said two toothed wheels.

[0040] The calculation of V c ( t j ) therefore uses values ​​of the respective rotational speeds of two toothed wheels in contact, in the epicyclic gear train. These values ​​used may belong to the primary rotation data, acquired in step 302. Alternatively, at least some of said values ​​used may be obtained by calculation, from the primary rotation data and numbers of teeth of the toothed wheels in the epicyclic gear train. Indeed, in an epicyclic gear train, the ratio of the rotational speeds of two toothed wheels in contact is a function of a ratio between the respective numbers of teeth of said wheels. In any event, the method according to the invention is preferably adapted to collect values ​​of rotational speeds relating to each of the three types of toothed wheel of the epicyclic gear train (crown 110, sun 120, satellite 130). These values ​​may all be acquired in step 302.Alternatively, some of these values ​​are not acquired in step 302, and are instead determined by calculation, using the above-mentioned relationship between the respective rotational speeds and number of teeth of two contacting gears in an epicyclic gear train.

[0041] Preferably, a value is calculated V c ( t j ) for every moment t j as defined above. The values V c ( t j ) thus calculated together form data called secondary rotation data.

[0042] In a fourth step 304, the primary mechanical wave data are used, S(t i ), and secondary rotation data, V c ( t j ), to simulate a sampling at constant phase intervals of the signal S formed by a progressive mechanical wave generated in the epicyclic gear train 100. Said phase is that of the contact point for which a rotation speed has been determined, in step 303. This step is detailed below, with reference to figures 4A à 4E . It preferably includes a temporal interpolation of the values S ( t i ) acquired in step 301 (primary mechanical wave data). At the end of step 304, a series of values ​​is obtained S (ϕ k ) of the signal S formed by the progressive mechanical wave generated in the epicyclic train 100, where ϕ k denotes a phase of the contact point considered in step 303. The values S ( ϕ k ) form data called secondary mechanical wave data.

[0043] Finally, in step 305, said secondary mechanical wave data are used to detect a possible anomaly in the operation of the epicyclic gear train. The information relating to the presence or absence of an anomaly in the operation of the epicyclic gear train, and where appropriate relating to the nature of a detected anomaly, together form so-called diagnostic information, D.

[0044] The anomaly may relate to a functional failure of the epicyclic gear train, to damage to its components, to operation outside nominal operating conditions (for example a cold start, or a lack of lubrication of the toothed wheels), or to any other cause likely to modify the progressive mechanical wave generated by the clashing of the teeth of the toothed wheels in contact in the epicyclic gear train 100. In particular, it is possible to detect a lack of lubrication of the planet gears, an increase in the torque on the planet gears, a lack of heat dissipation, a misalignment between the respective axes of rotation of at least two of the toothed wheels, etc. It is possible to detect sudden damage or progressive wear of the epicyclic gear train.

[0045] Anomaly detection uses secondary mechanical wave data, preferably transformed to facilitate analysis. Said transformed data is called analysis data. Said transformation aims in particular to remove noise and context effects. It preferably comprises the application of a Fourier transform. The Fourier transform is applied to the secondary mechanical wave data, or to a function obtained by interpolation of said secondary mechanical wave data. The Fourier transform here transforms a function of a phase into a function of an order. For example, a power spectral density, or an averaged power spectral density, or a spectrogram of the averaged power spectral densities, of the secondary mechanical wave data or of a function obtained by interpolation of said secondary mechanical wave data is calculated.

[0046] The detection of a possible anomaly in the operation of the epicyclic gear train can use models characterizing normal or abnormal operations of the epicyclic gear train, with identification of a deviation or similarity with respect to these models. These models are called health indicators.

[0047] The health indicators can be determined by numerical simulations, or using experimental measurements carried out on epicyclic gear trains associated with known operating characteristics and conditions. The method according to the invention can comprise preliminary steps for determining such health indicators, these preliminary steps comprising: for a plurality of flights of at least one aircraft, and where appropriate a plurality of flight phases of each flight, acquiring primary mechanical wave data and primary rotation data, and determining the corresponding secondary rotation data and secondary mechanical wave data; for each of the plurality of flights, and where appropriate for each of the plurality of flight phases of each flight, using the corresponding secondary mechanical wave data to calculate analysis data; and correlating said analysis data with information relating to the presence or absence of an anomaly linked to the operation of the epicyclic gear train, and where appropriate relating to the precise nature of said anomaly.

[0048] The flight phases include, in particular, one or more of a takeoff phase, a climb phase, a cruise phase, a descent phase, a landing phase, and a taxi phase. The analysis data correspond to the secondary mechanical wave data, transformed as described above.

[0049] Health indicators not only detect abnormal operation of the planetary gear train, but also identify the anomaly when it is detected. It is also possible to detect wear of the planetary gear train at an early stage, before significant defects appear. Early detection of wear of the planetary gear train offers great flexibility in planning a maintenance operation. This allows a maintenance operation schedule to be optimized, so as to minimize the total downtime of the aircraft receiving the planetary gear train.

[0050] In any event, the method may comprise a step, not shown, of maintenance on the epicyclic gear train 100, implemented in response to the detection of an anomaly, in step 305.

[0051] THE figures 4Aà 4E illustrate in detail a step 304 of phase-synchronous resampling.

[0052] There figure 4A illustrates the rotation speed of the contact point considered, V c ( t ), as a function of time t. The x-axis is a time, in seconds. The y-axis is a rotational speed, or rpm, in revolutions per minute. The curve V c ( t ) is obtained by interpolation of the secondary rotation data V c ( t j ). Here it corresponds to an acceleration phase.

[0053] There figure 4B illustrates the phase of the contact point considered, ϕ ( t), as a function of time t. The x-axis is a time, in seconds. The y-axis is a phase, in number of revolutions. The function ϕ ( t ) is obtained by integrating over time the function V c ( t ). The phase thus corresponds to a cumulative number of turns made by the point of contact considered.

[0054] There figure 4C corresponds to the figure 4B , on which points have been identified P n regular in-phase sampling. Each point P n is on the representative curve of ϕ ( t ) , and connects a value of the phase ϕ k and a value of time t ech_k . The different points P n are associated with regularly spaced phase values. In other words, the difference between the respective phases associated with two neighboring points P n is a constant, called the in-phase sampling step. Here, the in-phase sampling step is 1 revolution per minute.

[0055] There figure 4D illustrates the signal S formed by the progressive mechanical wave generated in the epicyclic train, as a function of time t. The abscissa axis is a time, in seconds. The ordinate axis here is an acoustic pressure, in Pa. The curve S(t) is obtained by interpolation of the primary mechanical wave data S ( t i ). On the figure 4D , sampling points have also been represented R n . The sampling points R n are points of the curve S(t) which correspond to the values t ech_k of the time t previously determined. These values t ech_k therefore correspond to regular in-phase sampling. Each sampling point R n associates a value S ( t ech_k ) of the signal S(t), and a value t ech_k of time t.

[0056] As illustrated in the figure 4C , each value t ech_k time is associated with a value ϕ k of the phase of the contact point considered. We can therefore link each value S ( t ech_k ) of the signal S(t) to a value ϕ k of the phase, where the sequence of values ϕ k corresponds to regular in-phase sampling. The values S ( t ech_k ) can therefore be noted S ( ϕ k ). They form the secondary mechanical wave data according to the invention, associated with regular sampling depending on the phase of the contact point considered.

[0057] On the figure 4E , the abscissa axis is a phase, in number of turns, and the ordinate axis is an acoustic pressure, in Pa. The points T n each associate one of the said values S ( ϕ k) with the corresponding phase value ϕ k . By interpolating the different points T n , we obtain a curve representative of a function S ( ϕ ). The function S ( ϕ ) represents the signal S formed by the progressive mechanical wave generated in the epicyclic train, as a function of the phase ϕ of the point of contact considered.

[0058] The transition from regular sampling as a function of time to regular sampling as a function of a phase makes it possible to overcome the effect of temporal variations in the rotation speeds of the toothed wheels of the epicyclic gear train. One of the clever ideas underlying the invention is to have chosen as the phase that of a contact point of the epicyclic gear train, rather than the phase of one of the toothed wheels of the latter. The inventors have shown that this choice allows for more relevant identification of anomalies linked to the operation of an epicyclic gear train.

[0059] As explained above, the detection of an anomaly linked to the operation of the epicyclic train is advantageously done on so-called analysis data, corresponding to the secondary mechanical wave data transformed to facilitate analysis.

[0060] The transformation preferably comprises the application of a function that involves a Fourier transform calculation. The Fourier transform here transforms a function of the phase ϕ of the point of contact considered, as a function of a variable called “order”.

[0061] The said function can be a power spectral density. The power spectral density is a function which represents the distribution in the orders of the power of the Fourier transform (square of the modulus of this Fourier transform). For a digital signal x n of N samples, an indicator of the power spectral density is a vector of N points, including thek - ème element is defined by: D X k = X k 2 N with D X the power spectral density of the signal x, And X ( k ) THE k - ème element of the discrete Fourier transform of the signal x.

[0062] Advantageously, the transformation of the secondary mechanical wave data to form the analysis data may comprise an interpolation of said secondary mechanical wave data. The function comprising a Fourier transform calculation is then applied to the result of said interpolation.

[0063] THE figures 5A, 5B And 6A, 6B illustrate the interest of carrying out regular resampling according to the phase of a point of contact then moving into the domain of orders.

[0064] There figure 5A illustrates the signal S formed by the progressive mechanical wave generated in the epicyclic train, as a function of time t. The abscissa axis is a time, in seconds. The ordinate axis here is an acoustic pressure, in Pa. The figure 5B illustrates the power spectral density of the signal from the figure 5A , as a function of a frequency in Hz.

[0065] There figure 6A illustrates the signal S formed by the progressive mechanical wave generated in the epicyclic train, as a function of the phase ϕ of the contact point considered. The abscissa axis is a phase, in number of turns, and the ordinate axis is an acoustic pressure, in Pa. The figure 6B illustrates the power spectral density of the signal from the figure 6A , depending on an order (inverse of a phase).

[0066] The signal of the figure 6A corresponds to a resampled signal, in which we are freed from the effect of temporal variations of the rotation speeds in the epicyclic gear train. By passing this signal into the domain of orders, we very easily identify a majority component centered on order 1 (see amplitude peak of the figure 6B ).

[0067] By resampling according to the phase of a contact point, followed by a passage in the order domain, one can easily discriminate between signal components due to the epicyclic train, and signal components due to noise and the environment around the epicyclic train.

[0068] The power spectral density values ​​corresponding to whole orders, and more particularly to order 1, are the most relevant for determining the presence or absence of an anomaly related to the operation of the epicyclic gear train, and if necessary, identifying the nature of the anomaly. Other values ​​of the orders can be used, preferably located at symmetrical positions on either side of a whole order.

[0069] The use of a power spectral density calculation to obtain the analysis data has been detailed above.

[0070] According to an advantageous variant, an averaged power spectral density (of the secondary mechanical wave data or of a function obtained by interpolating said secondary mechanical wave data) is instead calculated. The averaged power spectral density of a signal is obtained by multiplying this signal by an apodization window, calculating the power spectral density of the result of said multiplication, then repeating these steps after a shift of the apodization window by 50% of its length, and so on until a series of power spectral densities is obtained, the average of which is then calculated (Welch method). The apodization window is a function which takes the value zero everywhere except over a narrow band where it has a peak shape. It is defined for example by a Hamming function. This technique makes it possible to eliminate part of the noise and orders which are not dominant.

[0071] A spectrogram can also be used, grouping together several power spectral densities (averaged or not) calculated over successive time intervals. The time intervals considered overlap. They each correspond to a phase interval of the contact point considered. On the spectrogram, the power spectral density associated with a given phase interval corresponds to a vertical line. Said vertical line is aligned with a phase value which corresponds to the central value of said phase interval. An advantage of analyzing secondary mechanical wave data in the form of such a spectrogram is that the analysis is carried out on a compilation of numerous measurements, which makes it possible to overcome noise and marginal effects.

[0072] On the spectrogram, we can distinguish different remarkable lines, each corresponding to a high value of the power spectral density, which is maintained over time. Among these remarkable lines, only the straight lines, associated with constant values ​​of the order, relate to the epicyclic gear train. We can thus discriminate remarkable lines linked to the epicyclic gear train, and remarkable lines linked to the surrounding environment. This discrimination is particularly useful when the primary mechanical wave data are acoustic data, the acoustic radiation linked to the clashing of the teeth of the toothed wheels actually contributing only slightly to the total noise emitted at the level of a casing in which the epicyclic gear train is housed.

[0073] We therefore use spectrogram lines, or spectra associated with constant values ​​of the order, to determine the presence or absence of an anomaly related to the operation of the epicyclic gear train, and if necessary to identify the nature of the anomaly. These lines form acoustic signatures of the epicyclic gear train. We use in particular spectrogram lines associated with integer values ​​of the order, in particular the value one. The presence or absence of one or more of these lines forms an indicator of the health of the epicyclic gear train, characteristic of the presence or absence of an anomaly in its operation, and if necessary characteristic of the particular nature of the anomaly.

[0074] There figure 7 illustrates a detailed view of such a spectrogram, centered around order 1. The abscissa axis is a phase, in number of turns. The ordinate axis is an order. The hue scale is an amplitude of the power spectral density, in logarithmic scale. One of the power spectral densities forming the spectrogram is shown on the right, corresponding here to the vertical line AA' on the spectrogram. On the figure 7 , we have outlined in dotted lines an area with a horizontal line, aligned with order 1. This line indicates the presence of a maximum of the power spectral density, at order 1 and prolonged in time.

[0075] There figure 8 illustrates a detailed view of the same spectrogram, centered around order 2. On the right is shown one of the power spectral densities forming the spectrogram, corresponding here to the vertical line BB' on the spectrogram. A horizontal line is observed indicating the presence of a maximum of the power spectral density, at order 2 and prolonged in time (see boxed area).

[0076] Comparing the figures 7 et 8 , we observe an inversion of the preponderance of orders 1 and 2 (disappearance of the line at order 1 which coincides with the appearance of the line at order 2), which reflects an event on the epicyclic gear train. This event can be linked to a sudden degradation of the epicyclic gear train, or simply to a modification of the conditions of use, for example a modification of the torque applied to the epicyclic gear train. It is therefore advantageous to couple the analysis data calculated in the method according to the invention, with other measurements and monitoring carried out simultaneously in the aircraft (detection of impact of flying object, measurement of the quality of the oil lubricating the toothed wheels of the epicyclic gear train, detection of variations in torque, measurements of upstream and downstream temperatures, speed, position in the flight envelope, etc.).

[0077] There figure 9 schematically illustrates a system 10 according to the invention for implementing the method described with reference to the figure 3 . System 10 includes: an epicyclic gear train 100, as described with reference to figures 1, et 2 , which may be an integral part of a turboprop or a reducer; at least one mechanical wave sensor 11, for example an accelerometer, or more preferably an acoustic wave sensor such as a microphone, arranged to acquire a signal representative of a progressive mechanical wave generated in the epicyclic gear train 100 by the collision of the teeth of the toothed wheels (primary mechanical wave data); at least one rotation speed sensor 12, for acquiring rotation speed measurements relating to at least one of the toothed wheels of the epicyclic gear train 100 (primary rotation data); and a processor 13, configured to receive as input the primary mechanical wave data acquired by the mechanical wave sensor 11 and the primary rotation data acquired by the rotation speed sensor 12, to implement steps 303, 304 and 305 of the method illustrated in figure 3 , and to output diagnostic informationD relating to the presence or absence of an anomaly in the operation of the epicyclic gear train, and where applicable relating to the nature of the anomaly.

[0078] The mechanical wave sensor 11 is preferably located inside a housing, to protect it from dust or sand, and close to the epicyclic gear train. It is particularly advantageous to use an acoustic wave sensor as the mechanical wave sensor 11. Indeed, it is thus possible to acquire in a single signal, data relating to each of the toothed wheels making up the epicyclic gear train. In addition, it is thus possible to arrange the acoustic wave sensor remote from the epicyclic gear train, without direct physical contact with the latter. The acoustic wave sensor can even be located in an airport, to measure an acoustic signal emitted by an aircraft on approach or takeoff. The epicyclic gear train is then monitored from the noise it emits, in operation.

[0079] There figure 10 schematically illustrates a turbomachine of an aircraft receiving the epicyclic gear train 100. The latter is connected to an upstream propeller 20, a downstream propeller 30, and a turboshaft engine 40. The upstream and downstream propellers are counter-rotating (they turn in opposite directions) and unducted (they are located outside a fairing 50 receiving the epicyclic gear train 100 and the turboshaft engine 40). The upstream and downstream propellers have large dimensions, each with a diameter greater than 1.5 meters, for example two meters.

[0080] The invention is particularly suited to the field of high-power turbojets, for an aircraft, which see about 40% of the engine power pass through and which support energies of up to 30 MW / h. It applies in particular to the monitoring of a power transmission device with an epicyclic gear train, to transmit the power of a power turbine to upstream and downstream propellers which rotate in opposite directions.

[0081] The invention is based on a comparison between an expected model on the noise or vibrations emitted by an epicyclic train, and analysis data obtained by measurements and mathematical processing.

[0082] Alerts can be implemented to inform maintenance operators of the presence of an anomaly related to the operation of the epicyclic gear train. A maintenance recommendation service can also be set up for the aircraft operator.

[0083] Advantageously, the analysis data obtained using the secondary mechanical wave data according to the invention can be combined with additional data from accelerometer-type sensors.

[0084] In the examples detailed above, the example of an epicyclic gear train with three planet gears has been illustrated. The invention is not limited to this example, and can be implemented to monitor all kinds of epicyclic gear trains, with any number of planet gears, used as a reducer or as a propellant, or even several epicyclic gear trains arranged in series.

Claims

1. A method for monitoring an epicyclic gear train (100) of an aircraft, wherein the epicyclic gear train (100) includes two coaxially mounted first toothed wheels (110, 120) and at least one second toothed wheel (130) meshing with the first two toothed wheels, the method comprising the following steps of: - acquiring, at a predetermined sampling frequency, first values (S(ti)) of a signal formed by a progressive mechanical wave generated in the epicyclic gear train, said first values forming primary mechanical wave data; - measuring, at a plurality of successive time instants, values (Vmes_r(tj)) of a speed of rotation of at least one of the first two toothed wheels and the at least one second toothed wheel (110, 120, 130), said measured values forming primary rotation data; - from at least the primary mechanical wave data (S(ti)), determining second values (S(ϕk )) of said signal formed by a progressive mechanical wave generated in the epicyclic gear train, said second values corresponding to a regular sampling as a function of the phase and forming secondary mechanical wave data; and - using the secondary mechanical wave data (S(ϕk )) to detect an anomaly related to the operation of the epicyclic gear train (100), characterized it further comprises, the following step: - from primary rotation data (Vmes_r(tj)), calculating values (Vc(tj)) of a rotational speed of a point of contact between the first gear wheel (110, 120) and the second gear wheel (130) of the epicyclic gear train, said calculated values forming secondary rotation data ; and in that secondary mechanical wave data are determined using the primary mechanical wave data (S(ti))) and said secondary rotation data (Vc(tj)), and correspond to regular sampling as a function of the phase of said contact point.

2. The method according to claim 1, characterised in that the speed of rotation of the contact point between the one first toothed wheel (110; 120) and the one second toothed wheel (130) of the epicyclic gear train is a linear combination of the respective speeds of rotation of said first and second toothed wheels.

3. The method according to claim 2, characterised in that at least one of said respective speeds of rotation is calculated from the primary rotation data (Vmes(tj)) and the respective numbers of teeth of at least two of the first two toothed wheels and the at least one second toothed wheel (110, 120, 130).

4. The method according to any of claims 1 to 3, characterised in that using the secondary mechanical wave data (S(ϕk )) to detect an anomaly related to the operation of the epicyclic gear train (100) comprises: - calculating analysis data, including calculating a Fourier transform of the secondary mechanical wave data (S(ϕk )), where the Fourier transform transforms a function of a phase into a function of an order; and - using said analysis data to detect an anomaly related to the operation of the epicyclic gear train (100).

5. The method according to claim 4, characterised by using analysis data associated with at least one integer order, to detect an anomaly related to the operation of the epicyclic gear train (100).

6. The method according to claim 4 or 5, characterised in that calculating analysis data comprises calculating a power spectral density of the secondary mechanical wave data (S(ϕk )).

7. The method according to claim 4 or 5, characterised in that calculating analysis data comprises calculating an averaged power spectral density of the secondary mechanical wave data (S(ϕk )).

8. The method according to claim 4 or 5, characterised in that calculating analysis data comprises calculating a spectrogram gathering a plurality of values of an averaged power spectral density of the secondary mechanical wave data (S(ϕk )).

9. The method according to any of claims 1 to 8, characterised in that the signal formed by a progressive mechanical wave generated in the epicyclic gear train is an acoustic signal, acquired using an acoustic wave sensor (11).

10. The method according to claim 9, characterised in that the acoustic wave sensor (11) is arranged spaced from the epicyclic gear train (100).

11. The method according to any of claims 1 to 8, characterised in that the signal formed by a progressive mechanical wave generated in the epicyclic gear train is a vibration signal, acquired using an accelerometer.

12. The method according to any of claims 1 to 11, characterised in that it further includes a maintenance step on the epicyclic gear train (100), when an anomaly related to the operation of the latter is detected using the secondary mechanical wave data (S(ϕk )).

13. The method according to any of claims 1 to 12, characterised in that it further includes a preliminary step of determining health indicators, which includes the following steps of: - for a plurality of flights, acquiring primary mechanical wave data and primary rotation data, and determining the corresponding secondary rotation data and secondary mechanical wave data; - for each of the plurality of flights, calculating analysis data, said calculating including calculating a Fourier transform of the secondary mechanical wave data; and - correlating the analysis data with information relating to an anomaly or absence of anomaly in the operation of the epicyclic gear train (100).

14. A system (10) adapted to the implementation of a method according to any of the claims 1 to 13, the system comprising: - at least one mechanical wave sensor (11), arranged to acquire, at a predetermined sampling frequency, first values of a signal formed by a progressive mechanical wave generated in the epicyclic gear train, said first values forming primary mechanical wave data (S(ti)) ; - at least one speed of rotation sensor (12), arranged to acquire, at a plurality of successive time instants, values of a speed of rotation of at least one of the first two toothed wheels and the at least one second toothed wheel, said measured values forming primary rotation data (Vmes_r(tj)); and - a processor (13), configured to receive as an input the primary mechanical wave data (S(ti)) and the primary rotation data (Vmes_r(tj)), to determine second values (S(ϕk )) of said signal formed by a progressive mechanical wave generated in the epicyclic gear train, said second values being sampled as a function of a phase and named mechanical wave secondary data S(ϕk )), to determine from said mechanical wave secondary data S(ϕk )), an item of information (D) relating to the presence or absence of an epicyclic gear malfunction, and to output said information (D), characterized in that the processor (13) is further configured to : - calculate values (Vc(tj)) of a rotational speed of a point of contact between the first gear wheel (110, 120) and the second gear wheel (130) of the epicyclic gear train, said calculated values forming secondary rotation data (Vc(tj)); and - using the primary mechanical wave data (S(ti))) and said secondary secondary rotation data (Vc(tj)) to determine secondary mechanical wave data S(ϕk ), with said secondary mechanical wave data (S(ϕk)) corresponding to regular sampling as a function of the phase of said contact point.

15. The device (10) according to claim 14, characterised in that the mechanical wave sensor (11) is an acoustic sensor arranged spaced from the epicyclic gear train.