Method for determining a structural defect indicator

Abstract

A method is proposed for determining a structural defect indicator (I) of a component (10) of a running gear (20) of a land vehicle (100), the method comprising the steps: (i) acquiring, during a time window (W), a vibration signal (V) associated with the component (10) of the running gear (20), (ii) determining a sequence (S) of spectral components (A1, A2, …, Ap) of the vibration signal (V) acquired during the time window (W), (iii) iterating steps (i) and (ii) for a set of time windows (W1, …, Wn) so as to obtain a set of sequences (S1, ...(iv) determine a structural defect indicator (I) of the element (10) of the running gear (20) from the set of sequences (S1, …, Sn) of spectral components of the vibration signal (V) and from a detection model implementing a machine learning algorithm. Abstract figure: Figure 1.

Description

Title of the invention: Method for determining a structural defect indicator Technical field

[0001] The present invention relates to the field of methods for detecting mechanical defects in the ground connection elements of vehicles, such as trucks or trains. Prior art

[0002] Some vehicles have one or more axles on which the vehicle chassis rests via suspension springs, such as leaf springs. Sometimes, during the life of the vehicle, a spring may suffer damage caused by mechanical stress. Cracks may gradually spread into one or more leaves of the spring. The propagation of cracks can sometimes lead to total failure of the spring due to fatigue, which can create a dangerous situation and also immobilize the vehicle.

[0003] It is thus known to monitor the condition of the suspension springs during use of the vehicle, in order to detect damage before it can lead to a failure. To do this, it is known to measure the acceleration of the spring itself, or of the axle. Comparing the measured acceleration with a threshold makes it possible to diagnose damage to the spring. However, due in particular to variations in the conditions of use, such as the load of the vehicles, their speed or the profile of the road, the known methods are not satisfactory. Indeed, the known methods often have a non-detection rate and a false detection rate that are too high. In addition, most of the known methods only allow a qualitative diagnosis, and not a quantitative diagnosis. In other words, the presence of a defect can be detected, but the severity of the detected defect remains difficult to evaluate.

[0004] The present disclosure aims to improve this situation by proposing an improved method for establishing the presence of a defect in a more precise and robust manner, so as to be able to anticipate corrective action. Summary

[0005] To this end, the invention proposes a method for determining a structural defect indicator of an element of a land vehicle running gear, the method comprising the steps: - (i) acquire during a time window a vibration signal associated with the element of the running gear, - (ii) determine a sequence of spectral components of the acquired vibration signal during the time window, - (iii) iterating steps (i) and (ii) for a set of time windows so as to obtain a set of sequences of spectral components of the vibration signal, each sequence of the set of sequences corresponding to a time window, - (iv) determine a structural defect indicator of the running gear element from the set of sequences of spectral components of the vibration signal and from a detection model implementing a machine learning algorithm.

[0006] The features listed in the following paragraphs can be implemented independently of each other, or in any technically possible combination:

[0007] According to one embodiment of the method, the spectral components of the sequence of spectral components are determined by fast Fourier transform of the samples of the vibration signal acquired during the time window.

[0008] Each spectral component of a sequence of spectral components of the vibration signal is thus a value calculated over the entire corresponding acquisition time window.

[0009] According to an example of implementation of the method, the structural defect is a crack in the element of the running gear.

[0010] According to an example of implementation of the method, the element of the running gear is an elastic suspension element of the vehicle.

[0011] The elastic suspension element is for example a leaf spring.

[0012] The leaf spring is disposed between an axle of the vehicle and a chassis of the vehicle.

[0013] Alternatively, the elastic suspension element may be a helical spring.

[0014] Also as a variant, the elastic suspension element may be a torsion bar.

[0015] As a further variant, the elastic suspension element may be an air cushion.

[0016] The chassis of the vehicle may be a self-supporting body of the vehicle.

[0017] According to an example of application of the method, the structural defect is a defect in a bearing of a wheel connected to the axle.

[0018] A structural defect such as damage to a ball or part of the raceway of one of the bearing rings modifies the vibration response of the running gear and can be detected by analysis of the acquired vibration signal.

[0019] According to another example of application of the method, the structural defect is a structural defect of a tire of a wheel connected to the axle.

[0020] As previously, a structural defect such as a deformation of a tire of the running gear modifies the vibration response of the running gear and can be detected by analysis of the acquired vibration signal.

[0021] According to yet another example of application of the method, the structural defect is an overload of the axle.

[0022] Similarly, excessive weight carried by the axle can be detected by analyzing the acquired vibration signal.

[0023] According to yet another application of the method, the structural defect is an excessive reduction in the damping coefficient of the relative movements of the axle with respect to the chassis of the vehicle. Wear of the axle oscillation dampers can also be detected by analyzing the acquired vibration signal.

[0024] In an example application of the method, the vehicle is a truck, or a van.

[0025] In another example of application of the method, the vehicle is railway rolling stock, such as a wagon or a train locomotive.

[0026] According to one embodiment of the method, the vibration signal is an acceleration signal from an acceleration sensor.

[0027] The acceleration sensor is for example arranged on the element of the vehicle's running gear.

[0028] In other words, the acceleration sensor is then placed on the element whose operating state is to be monitored.

[0029] According to another implementation of the method, the acceleration sensor is arranged on the axle.

[0030] According to yet another implementation of the method, the acceleration sensor is arranged on the chassis of the vehicle.

[0031] In these two implementation examples, the acceleration sensor is arranged on an element other than the element to be monitored, while being mechanically linked to the element to be monitored. The installation of the acceleration sensor is thus facilitated since the number of installation possibilities is increased.

[0032] The method can jointly use the signal from several acceleration sensors arranged on different elements of the running gear or on the chassis.

[0033] According to one embodiment, the acceleration sensor is a three-axis sensor.

[0034] Alternatively, the acceleration sensor may be a single-axis sensor.

[0035] The sampling frequency is greater than 30 Hz. According to one embodiment, the acquisition frequency, also called sampling frequency, of the vibration signal from the accelerometer is between 25 kHz and 35 kHz, for example equal to 32 kHz.

[0036] According to one embodiment of the method, each sequence of the set of sequences of spectral components of the vibration signal is obtained by averaging a plurality of sequences of spectral components corresponding to a plurality of windows of elementary acquisitions of the vibration signal.

[0037] The plurality of elementary acquisition windows may comprise between 5 acquisition windows and 100 acquisition windows.

[0038] The duration of a time window W for acquiring the vibration signal V is configurable. The duration of a time window for acquiring the vibration signal is for example between 2 seconds and 200 seconds. The duration of a time window for acquiring the vibration signal may be equal to 100 seconds.

[0039] The time windows for acquiring the vibration signal may be contiguous.

[0040] According to one embodiment, the method comprises a step: - (üil) determine the vehicle's forward speed, - (iii2) eliminate from the set of acquired time windows the time windows corresponding to a vehicle forward speed lower than a predetermined minimum threshold.

[0041] The elimination of these acquisition windows makes it possible to improve the signal-to-noise ratio, because these acquisition windows correspond to phases of operation of the vehicle during which the vibration signal is weak.

[0042] The method may comprise a step: - (üi3) eliminate from the set of acquired time windows the time windows corresponding to zero vehicle speed.

[0043] Alternatively or additionally, the method comprises a step: - (iii2') eliminate from the set of acquired time windows the time windows corresponding to a vehicle forward speed greater than a predetermined maximum threshold.

[0044] The elimination of these acquisition windows makes it possible to avoid possible saturations due to a high signal level.

[0045] The method may comprise a step of normalizing the data. This means that the scale of the signals is rescaled so as to have a zero mean and a standard deviation equal to 1. The data used for the rest of the process steps are therefore those of the set of normalized spectral component sequences.

[0046] Step (iv) of determining a structural defect indicator method comprises a step (ivl) of reducing the dimensions of the set of sequences of spectral components of the vibration signal, so as to obtain a set of reduced sequences of combinations of spectral components of the vibration signal.

[0047] Each reduced sequence comprises a number of spectral components less than the number of spectral components determined in step (ii).

[0048] The dimension reduction step makes it possible to reduce the complexity of the calculations carried out, while retaining the components which best characterize the data. acquired.

[0049] According to one embodiment of the method, the step (ivl) of reducing the dimensions of the set of sequences of spectral components of the vibration signal implements an unsupervised dimension reduction algorithm.

[0050] An unsupervised algorithm allows for simple adaptation to a wide variety of technical definitions. Indeed, an unsupervised machine learning algorithm allows for the construction of mathematical models based on actually observed data, which here are the acquired samples of the vibration signal.

[0051] Different unsupervised dimension reduction algorithms can be used.

[0052] For example, an autoencoder network may be used. The step (ivl) of reducing the dimensions of the set of sequences of spectral components of the vibration signal is thus carried out by encoding a sequence of spectral components of the vibration signal, using an encoder of a pre-trained autoencoder network, in a latent space of reduced dimension.

[0053] A principal component analysis algorithm is another example of a usable algorithm.

[0054] The unsupervised dimension reduction algorithm generates a dimension reduction model.

[0055] The set of reduced sequences of spectral components can thus be determined by a model obtained by machine learning.

[0056] The method can thus automatically adapt to the mechanical configuration on which the method is used.

[0057] According to an example of implementation of the method, a reduced sequence of spectral components comprises for example seven elements.

[0058] According to another example of implementation of the method, a reduced sequence of spectral components comprises for example three elements.

[0059] According to yet another example of implementation of the method, a reduced sequence of spectral components comprises for example two elements.

[0060] The associated calculations can thus be reduced by a large factor, compared to the data initially acquired.

[0061] The dimension reduction model of the set of sequences of spectral components of the vibration signal can be obtained by automatic learning of the vibration signal from a subset of time windows acquired under reference conditions in which the element of the running gear is free from structural defects.

[0062] The method comprises a step: - (iv2) determine, for each time window, a reconstructed sequence of com spectral components of the vibration signal acquired during the time window from the reduced sequence of combinations of spectral components and from a reconstruction model obtained from the dimension reduction model.

[0063] The model for reconstructing the spectral components of the acquired vibration signal is thus based on automatic learning of the vibration signal from a subset of time windows acquired under reference conditions in which the element of the running gear is free from structural defects.

[0064] According to one aspect of the method, the reconstruction model of spectral components of the acquired vibration signal is obtained by inversion of the dimension reduction model.

[0065] The reconstruction model can thus be easily determined from the model already created for the reduction of dimensions of the acquired data.

[0066] According to one aspect of the disclosure, the method comprises a step: - (iv3) Determine, for each time window, a difference between the sequence of spectral components of the acquired vibration signal and the reconstructed sequence of spectral components of the vibration signal, and the structural defect indicator is based on the determined deviation.

[0067] In the absence of a structural defect, the reconstructed signal is close to the initially acquired signal. A relatively larger deviation indicates the behavior of the mechanical system deviates from the modeled behavior in the absence of a defect, which indicates the presence of a defect.

[0068] According to one embodiment of the method, the difference between the series of spectral components of the acquired vibration signal and the reconstructed series of spectral components of the vibration signal is determined for all of the spectral components of the vibration signal.

[0069] In this embodiment, the spectral components of the signal are reconstructed for all of the spectral frequencies, and the difference between the original signal and the reconstructed signal is also determined by taking into account all of the spectral frequencies, i.e. all of the frequency information of the signal.

[0070] According to one embodiment of the method, the difference between the series of spectral components of the acquired vibration signal and the reconstructed series of spectral components of the vibration signal is determined for a subset of spectral components. The subset of spectral components is included in the suite of spectral components of the acquired vibration signal.

[0071] In other words, in this case the spectral components of the signal are reconstructed for all spectral frequencies, but the deviation between the original signal and the reconstructed signal is determined by taking into account only certain selected frequency ranges. The deviation can thus be quantified only in the ranges of frequency most favorable to the highlighting of a structural defect.

[0072] According to an exemplary embodiment of the method, the determined difference is the Euclidean distance between the sequence of spectral components of the acquired vibration signal and the reconstructed sequence of spectral components of the vibration signal.

[0073] According to one aspect of the disclosure, the method comprises a step: - (iv4) Determine a detection threshold, the detection threshold being equal to the maximum value of the deviation determined for a subset of time windows acquired under reference conditions in which the running gear element is free from structural defects.

[0074] The learning carried out under reference conditions makes it possible to define the threshold defining the defect. The method can thus be automatically adapted to any mechanical configuration. In addition, the threshold for defining a defect can be periodically updated as data acquired under reference conditions is collected. The robustness of the method is thus improved.

[0075] According to an exemplary implementation, the method comprises a step: - (iv5-a) if the determined deviation is less than or equal to the detection threshold, the structural defect indicator takes a first value corresponding to an absence of structural defect, - (iv5-b) if the determined deviation is greater than the detection threshold, the structural defect indicator takes a second value corresponding to the presence of a structural defect.

[0076] A large deviation between the acquired vibration signal and the reconstructed vibration signal tends to indicate the presence of a structural defect in the mechanical system. A small deviation indicates that the system response is consistent with the behavior in the absence of a defect.

[0077] The structural defect indicator can take discrete values, a first value corresponding to an absence of structural defect and a second value corresponding to a presence of structural defect.

[0078] According to one embodiment, the method comprises a step: - (v) emit an alert signal in response based on the structural defect indicator.

[0079] The alert signal emitted makes it possible to trigger an in-depth analysis of the vehicle component and to anticipate a failure.

[0080] The method can thus comprise the steps: - (vl) determine an average value of the structural defect indicator, - (v2) emit an alert signal if a deviation between the determined average value and a reference value is greater than a predetermined threshold.

[0081] An isolated change in the value of the structural fault indicator does not trigger warning signal. Repeated changes in values ​​trigger the emission of a warning signal.

[0082] The alert signal may be a fault code stored in an electronic control unit.

[0083] According to a variant or in a complementary manner, the alert signal may be the lighting of a warning light.

[0084] According to another variant or also in a complementary manner, the alert signal can be a message sent to a vehicle manager.

[0085] The disclosure also relates to a device for detecting a structural defect in an element of a land vehicle running gear, comprising: - at least one accelerometer, - a device for acquiring the accelerometer signal, - an electronic control unit configured to implement the method described above.

[0086] According to one embodiment of the detection device, the electronic control unit is external to the vehicle, and the detection device comprises a communication means configured to transmit the acquisitions or the spectral components of the signal from the accelerometer to the electronic control unit.

[0087] The detection device thus comprises an on-board calculation means configured to determine the set of sequences of spectral components of the vibration signal. On-board calculation means is understood to mean a calculation means present on the vehicle. The signal transmitted to the control unit external to the vehicle may be a frequency signal or a raw time signal. Transmitting a frequency signal makes it possible to limit the quantity of data exchanged. Alternatively, the raw time signal and the frequency signal may both be transmitted. Brief description of the drawings

[0088] Other characteristics, details and advantages will appear on reading the detailed description below, and on analyzing the attached drawings, in which:

[0089] [Fig.l] is a schematic representation of a vehicle comprising a device for detecting a structural defect in an element of a running gear,

[0090] [Fig.2] is a detail of the running gear of the vehicle of [Fig.l],

[0091] [Fig.3] is a schematic time curve illustrating the acquisition step of vibration signal,

[0092] [Fig.4] is a curve illustrating the temporal evolution of the acquired vibration signal,

[0093] [Fig.5] is a view illustrating a particular step of the method,

[0094] [Fig.6] is a diagram illustrating certain steps of the process,

[0095] [Fig.7] is a view illustrating another particular step of the method,

[0096] [Fig.8] is a view illustrating another particular step of the method,

[0097] [Fig.9] is a block diagram illustrating different steps of the method according to the invention. Description of the embodiments

[0098] In order to facilitate reading of the figures, the different elements are not necessarily represented to scale. In these figures, identical elements bear the same references. Certain elements or parameters may be indexed, that is to say designated for example by first element or second element, or first parameter and second parameter, etc. This indexing is intended to differentiate similar, but not identical, elements or parameters. This indexing does not imply a priority of one element or parameter over another and the names may be interchanged. When it is specified that a device comprises a given element, this does not exclude the presence of other elements in this device.

[0099] [Fig.l] shows a land vehicle 100 comprising a running gear 20. The running gear 20 designates all of the mechanical components participating in the connection between the chassis 2 of the vehicle and the ground on which the vehicle moves.

[0100] The vehicle 100 is here a truck. The running gear 20 comprises an axle 1 connected to the chassis 2 by an elastic element 10. The elastic element is here a leaf spring, or more precisely a leaf spring arranged near each of the ends of the axle 1. The axle 1 comprises at each of its ends a wheel 4 connected to the axle by a bearing 3. The wheel 4 is equipped with a tire 5.

[0101] It is common for an element of the running gear 20 to experience a structural defect during the life of the vehicle. A structural defect is understood to mean a mechanical anomaly affecting the mechanical properties of the member in question. Such a structural defect may, as it develops, lead to a failure of the running gear, which may degrade the operation of the vehicle and possibly render it inoperable. Even if the running gears are generally inspected during the maintenance phases of the vehicle, it is desirable to be able to detect a possible structural defect of one of the elements of the running gear during use of the vehicle, in order to be able to remedy this defect and anticipate possible failures.

[0102] A method is thus proposed for determining an indicator I of a structural defect of an element 10 of a running gear 20 of a land vehicle 100, the method comprising the steps: - (i) acquiring during a time window W a vibration signal V associated with the element 10 of the running gear 20, - (ii) determine a sequence S of spectral components (Al, A2, ..., Ap) of the vibrational signal V acquired during the time window W, - (iii) iterate steps (i) and (ii) for a set of time windows Wb ..Wn so as to obtain a set of sequences Sb Sn of spectral components of the vibration signal V, each sequence Sk of the set of sequences Sb Sn corresponding to a time window Wk, - (iv) determining an indicator I of structural defect of the element 10 of the running gear 20 from the set of sequences Sb Sn of spectral components of the vibration signal V and from a detection model implementing a machine learning algorithm.

[0103] According to an example of implementation of the method, the structural defect is a crack in the element 10 of the running gear 20.

[0104] The element 10 of the running gear 20 is here an elastic suspension element of the vehicle 100.

[0105] More specifically, the elastic suspension element 10 is here a leaf spring. The leaf spring 10 is arranged between an axle 1 of the vehicle and a chassis 2 of the vehicle.

[0106] According to a non-illustrated example, the elastic suspension element 10 may be a helical spring. According to yet another non-illustrated example, the elastic suspension element 10 is a torsion bar. According to another non-illustrated example, the elastic suspension element 10 is an air cushion.

[0107] In the illustrated example, the chassis 2 comprises two parallel side members connected by transverse cross members, and a cabin rests on the chassis 2. According to an example not illustrated, the chassis 2 of the vehicle 100 may be a self-supporting body of the vehicle 100.

[0108] Various types of structural defects can be detected by the proposed method. The structural defect may in particular be a defect in a bearing 3 of a wheel 4 connected to the axle 1. The structural defect may also be a structural defect in a tire 5 of a wheel 4 connected to the axle 1. The structural defect may also be an overload of the axle 1. The structural defect may also be wear of the dampers of the oscillations of the axle.

[0109] When one of these structural defects is present, the vibration response of the running gear 20 is modified. It is thus possible to detect the presence of a structural defect by analyzing the acquired vibration signal. The frequency range in which the signal is used varies depending on the nature of the defect and the targeted application.

[0110] The vehicle 100 is here a truck. The vehicle can also be a van. In another example of application of the method, the vehicle 100 is railway rolling stock. The vehicle 100 can thus be a wagon or a train locomotive. In this case, the cabin of the wagon, or of the locomotive, rests on a series of bogies suspended by a set of elastic elements such as leaf springs.

[0111] The vibration signal V is an acceleration signal from an acceleration sensor 7.

[0112] The acceleration sensor 7 can be installed on different elements or organs. The acceleration sensor 7 is rigidly fixed, for example by gluing, to the outer surface of a member linked to the running gear 20.

[0113] The acceleration sensor 7 can be arranged on the element 10 of the running gear 20 of the vehicle. In the example illustrated in [Fig.2], an acceleration sensor 7 is arranged on the leaf spring 10. The acceleration sensor 7 is thus integral with the element whose operating state is to be monitored.

[0114] According to the example shown schematically in [Fig.l], the acceleration sensor 7 is arranged on the axle 1. According to an example not shown, the acceleration sensor 7 is arranged on the chassis 2 of the vehicle 100.

[0115] In these two configurations, the acceleration sensor 7 is arranged on an element other than the element to be monitored. The installation of the acceleration sensor 7 is thus facilitated, since it can then be installed in a more accessible location and less exposed to possible projections from the wheels, such as for example the chassis of the vehicle.

[0116] The method can use in parallel several signals acquired simultaneously by several acceleration sensors arranged on different elements of the running gear or on the chassis. For example, it is possible to use an acceleration sensor arranged on a leaf spring and another acceleration sensor arranged on the chassis. It is also possible to use an acceleration sensor arranged on a leaf spring and an acceleration sensor arranged on the corresponding axle. It is also possible to jointly use a third acceleration sensor arranged on the chassis.

[0117] It is said that the vibration signal V is associated with the element 10 of the running gear 20, because the vibration signal V reflects the vibrations within the element 10 of the running gear 20, even when fixed to a mechanical member other than the element 10, such as the axle or the chassis of the vehicle.

[0118] The acceleration sensor 7 is for example a three-axis sensor. The sensor delivers a signal dependent on the acceleration along the three directions in space. In Figures 1 and 2, the Z direction represents the vertical axis, the X direction represents the longitudinal axis of the vehicle and the Y direction represents the transverse direction of the vehicle. A three-axis sensor is thus sensitive to accelerations along each of the X, Y, Z axes.

[0119] Alternatively, the acceleration sensor 7 may be a single-axis sensor. In this case, the sensor 7 delivers a signal related to the acceleration in a single direction. Preferably, the sensor is installed so as to be preferentially sensitive to the acceleration along the vertical axis Z.

[0120] The acceleration sensor 7 delivers an analog signal. The bandwidth of the acceleration sensor 7 is greater than 30 Hz. For example, the bandwidth of the acceleration sensor 7 is between 25 kHz and 35 kHz.

[0121] An acquisition device 8 acquires the signal V from the accelerometer 7. During an acquisition time window W, the vibration signal V is sampled at a predetermined acquisition frequency Fa. The acquisition frequency Fa of the vibration signal V is preferably constant. In the context of the present disclosure, acquisition frequency and sampling frequency are equivalent terms.

[0122] The acquisition frequency Fa, also called sampling frequency, of the vibration signal V of the accelerometer 7 is greater than 30 Hz. The acquisition frequency Fa can be between 25 kHz and 35 kHz. This sampling frequency is for example equal to 32 kHz.

[0123] [Fig. 3] schematically illustrates the acquisition time windows. A plurality of successive acquisition windows, designated by increasing indices from Wk to Wk+e are represented. The arrows designated by the sign g schematically illustrate the sampling instants. The sampling instants are separated by a duration Ta equal to the inverse of the sampling frequency Fa. To simplify the figure, the sampling instants have been represented only for the Wk time acquisition window of index k, and have not been represented for the other windows.

[0124] In the example of [Fig.3], the time windows for acquiring the vibration signal V are contiguous. In other words, the start of an acquisition time window Wj corresponds to the end of the previous acquisition window Wj.i.

[0125] According to an example not illustrated, the time windows for acquiring the vibration signal V can be disjoint, that is to say that an acquisition time window Wj begins later than the end of the previous acquisition window Wj i.

[0126] [Fig.4] illustrates the shape of the vibration signal V during a total acquisition time of 30 minutes. The peaks included for example between times ti and t2 correspond to operating phases in which the accelerations are the highest. Many other peaks are visible, for example around the time designated by t3.

[0127] The proposed method uses the characteristics of the acceleration signal in the frequency domain, and therefore determines the spectral components of the acquired vibration signal V. The set of spectral components (Al, A2, ..., Ap) of the signal acquired during a given acquisition window is designated by the term 'sequence' S. For an acquisition time window of index k, designated by Wk, the corresponding sequence is designated by Sk, i.e. with the same index k.

[0128] The spectral components (Al, A2, ..., Ap) of the sequence S of spectral components are for example determined by fast Fourier transform of the samples of the vibration signal V acquired during time window W.

[0129] Once the vibration signal acquisitions have been carried out, the rest of the process steps can be carried out in real time, or in deferred processing.

[0130] Each spectral component (Al, A2, ..., Ap) of a sequence Sk of spectral components of the vibration signal V is thus a value calculated over the entire corresponding acquisition time window Wk. The components (Al, A2, ..., Ap) of a sequence Sk of spectral components are for example calculated in decibels.

[0131] Each component Ai of a sequence Sk of spectral components corresponds to a frequency Fi. The component designated by Ai is the i-th spectral component of the sequence Sk. The range of frequencies Fi extends for example from 0.5 Hz to 16 kHz.

[0132] The set of sequences Sb ..., Sn of spectral components is presented as a two-dimensional matrix. The columns can correspond to the different spectral components (Al, A2, ..., Ap), arranged in ascending order. The matrix rows then correspond to the different acquisition time windows Wi, ..., Wn, arranged in chronological order of acquisition. It is of course possible to arrange the different spectral frequencies in rows, and the different time samples in columns, that is to say to transpose the matrix.

[0133] The duration of a time window W for acquiring the vibration signal V is configurable. The duration of a time window W for acquiring the vibration signal V can be between 2 seconds and 200 seconds. The duration of a time window W for acquiring the vibration signal V is for example equal to 100 seconds.

[0134] According to one embodiment of the method, each sequence Sk of the set of sequences Sb ..., Sn of spectral components of the vibration signal V is obtained by averaging a plurality of sequences of spectral components corresponding to a plurality of elementary acquisition windows of the vibration signal V.

[0135] The plurality of elementary acquisition windows may comprise, for example, between 5 acquisition windows and 100 acquisition windows.

[0136] In other words, rather than using an acquisition window W of a given duration, it is also possible to use several acquisition windows of shorter duration, and to average the results obtained over each of these windows. For example, rather than acquiring the vibration signal V during a time window lasting 100 seconds, it is possible to acquire 50 successive windows each lasting 2 seconds, and to average the results obtained over the 50 windows, also representing 100 seconds of acquisition in total. This averaging step is optional, and the data processing carried out subsequently is identical.

[0137] According to one embodiment, the method comprises a step: - (üil) determining the forward speed C of the vehicle 100, - (iii2) eliminating from the set of acquired time windows Wi, ..., Wn the time windows corresponding to a forward speed C of the vehicle 100 lower than a predetermined minimum threshold Cmin.

[0138] Eliminating acquisition windows acquired while the vehicle speed is below the minimum threshold Cmin makes it possible to improve the signal-to-noise ratio. Indeed, these acquisition windows correspond to phases of vehicle operation during which the vibration signal V is weak, and therefore not very rich in relevant information. In the example of [Fig.4], it can be seen that the signal is very weak during the operating periods between times t4 and t5, as well as for the periods between times t6 and t7. These periods correspond to phases during which the vehicle is traveling at very low speed, or is stationary. It is thus possible to eliminate from the acquired results the points corresponding to these time periods.

[0139] For a road vehicle, the threshold Cmin is for example equal to 10 km / h. When the forward speed C of the vehicle 100 is lower than the predetermined threshold Cmin for the entire duration of an acquisition window Wi of the vibration signal V, this acquisition window Wi is eliminated from all of the acquired data. In other words, the data acquired during this acquisition window Wi are not used by the following steps of the method.

[0140] For example, the method may comprise a step: - (üi3) eliminate from the set of acquired time windows Wi, ..., Wn the time windows corresponding to a zero speed of the vehicle 100. In other words, only the time windows corresponding to a rolling vehicle are kept.

[0141] Given the acquisition frequency and duration, the matrix formed by the set of sequences Si, ..., Sn of spectral components (Al, A2, ..., Ap) of the vibration signal V has a large number of rows as well as columns. This matrix can comprise several thousand spectral components and several hundred time windows. However, the majority of the variance of the calculated data can be preserved by using significantly fewer independent spectral frequencies. The calculations can thus be significantly simplified while maintaining sufficient accuracy.

[0142] Alternatively or additionally, the method comprises a step: - (iii2') eliminate from the set of acquired time windows (Wb ..., Wn) the time windows corresponding to a forward speed C of the vehicle 100 greater than a predetermined maximum threshold Cmax. This step can help eliminate acquisition windows where signal saturation may occur.

[0143] The method may comprise a step of normalizing the data. By where the scale of the signals is rescaled so as to have a zero mean and a standard deviation equal to 1. The rescale transformation here is an affine transformation. For this, the mean and the standard deviation of the spectral components (Al, A2, ..., Ap) are determined for each sequence Sk of the set of sequences Si, ..., Sn. Then, each sample is normalized by subtracting the previously determined mean and dividing the result of this subtraction by the previously determined standard deviation.

[0144] The normalized value Xinorm of a sample X; of a variable X having a mean Avg and a standard deviation Ec is thus:

[0145] [Math.l] Y _ X, - Moy i_norm Ec

[0146] The data used for the rest of the steps of the method are thus the normalized values ​​of the set of sequences (Si, ..., Sn) of spectral components (Al, A2, ..., Ap). In the example of [Fig.5], the dimension reduction down to 2 dimensions was carried out on a normalized data set. The two principal components thus have values ​​distributed on either side of the zero value. The normalization step is optional.

[0147] Step (iv) of determining a structural defect indicator I of the proposed method thus comprises a step (ivl) of reducing the dimensions of the set of sequences Si, ..., Sn of spectral components (Al, A2, ..., Ap) of the vibration signal V, so as to obtain a set of reduced sequences SRi, ..., SRn of combinations of spectral components of the vibration signal V. Each reduced sequence SRk comprises a number q of spectral components less than the number p of spectral components determined in step (ii). All the reduced sequences have the same number q of spectral frequencies.

[0148] The dimension reduction step makes it possible to reduce the complexity of the calculations carried out. The number of columns of the matrix forming the set of sequences of spectral components is thus reduced compared to the initial data constituted. The number of rows of the matrix remains unchanged, because all the acquisition windows are preserved.

[0149] According to one embodiment of the method, the step (ivl) of dimension reduction of the set of sequences Si, ..., Sn of spectral components of the vibration signal V implements an unsupervised dimension reduction algorithm. An unsupervised algorithm allows simple adaptation to a wide variety of technical definitions.

[0150] The algorithm implemented for dimension reduction can be based on an auto-encoder network, commonly referred to by the English acronym “DAE” for “Deep Auto Encoder”. Step (iv 1) of reducing the dimensions of the set of sequences of spectral components of the vibration signal is thus carried out by encoding a sequence of spectral components of the vibration signal, using an encoder of a pre-trained auto-encoder network, in a latent space of reduced dimension.

[0151] A principal component analysis algorithm, commonly referred to by the English acronym “PCA” for “Principal Component Analysis” is another example of an algorithm that can be implemented to carry out the dimension reduction step.

[0152] The unsupervised dimension reduction algorithm generates a dimension reduction model M.

[0153] The set of reduced sequences SRi, ..., SRn of spectral components is thus determined by a model M obtained by machine learning.

[0154] The method can thus automatically adapt to the mechanical configuration on which the method is used.

[0155] According to an example of implementation of the method, a reduced sequence SRk of spectral components comprises for example seven elements. In this case, a reduced sequence SRk of spectral components comprises seven frequency elements. The number q is 7. According to another example of implementation of the method, a reduced sequence SRk of spectral components comprises for example three elements. The number q is thus 3. According to yet another example of implementation of the method, a reduced sequence SRk of spectral components comprises, for example, two elements. The number q is then equal to 2.

[0156] In other words, the matrix forming the set of sequences of spectral components then comprises only 7 columns, 3 columns, 2 columns respectively. The associated calculations can thus be reduced by a large factor compared to the initially acquired data. The matrix comprises as many rows as there are time windows retained after the steps of eliminating the windows acquired under unfavorable conditions, for example for vehicle speeds that are too low or too high, as seen previously.

[0157] The model M for reducing the dimensions of the set of sequences (Si, ..., Sn) of spectral components of the vibration signal V is obtained by automatic learning of the vibration signal V from a subset of time windows acquired under reference conditions in which the element 10 of the running gear 20 is free from structural defects. For example, the autoencoder network is trained under conditions in which a structural defect is not present.

[0158] Detailed checks of the mechanical condition of the various elements of the train rolling 20 are preferably carried out by one or more maintenance operators before the reference data acquisition phase. When a detailed check is not desired because it is too costly in terms of resources, the new condition of the vehicle can be considered as representative of the reference conditions without structural defects and defining the nominal operating condition of the vehicle.

[0159] [Fig.5] illustrates results obtained after a dimension reduction down to 2 dimensions. The results are thus plottable in a plane, each axis representing a linear combination of spectral frequencies. Each point of the point cloud corresponds to a time acquisition window. The horizontal axis Cl corresponds to a first combination of spectral frequencies and the vertical axis C2 corresponds to a second combination of spectral frequencies. As previously indicated, each acquisition window may come from an average of a plurality of elementary acquisition windows and certain elementary windows may have been eliminated because they correspond to too low vehicle speeds. For convenience, examples where dimension reduction is done to three or seven dimensions have not been shown.

[0160] The next step of the method consists of reconstructing the initially acquired vibration signal, or more precisely the spectral components of the acquired vibration signal, from the reduced sequence of determined spectral components.

[0161] For this, the method comprises a step: - (iv2) determine, for each time window Wk, a reconstructed sequence S'k of spectral components A' 1, A'2, ..., A'p of the vibration signal V acquired during the time window Wk from the reduced sequence SRk of combinations of spectral components and from a reconstruction model P obtained from the dimension reduction model M.

[0162] The step of reconstructing the acquired vibration signal is the inverse step of the dimension reduction step. [Fig.6] schematically illustrates the organization of the different steps of the method, with a series of data processing steps aimed at achieving a dimension reduction of the data from the acquired signal and a series of inverse processing, making it possible to compare the acquired signal and a reconstructed signal.

[0163] The reconstruction model P of the spectral components A' 1, A'2, ..., A'p of the acquired vibration signal V is thus based on automatic learning of the vibration signal V from a subset of time windows acquired under reference conditions in which the element 10 of the running gear 20 is free from structural defects.

[0164] The reconstruction model P of the acquired vibration signal V is obtained by inversion of the dimension reduction model M.

[0165] The reconstruction model can thus be easily determined from the model already created for the reduction of dimensions of the acquired data.

[0166] When the dimension reduction algorithm of the original matrix is ​​of the principal component analysis type, the dimension reduction step amounts to projecting the multidimensional space formed by the original data onto principal directions of a reduced-dimensional space. For this, the original data matrix Y is decomposed into a product of three matrices of the form B x C x DT, matrices B and D being orthogonal matrices, therefore invertible. DT denotes the transposed matrix of D. Matrix D contains the eigenvectors of the covariance matrix of Y. Dimension reduction is achieved by multiplying the original matrix Y by a projection matrix D' containing a selection of the columns of matrix D corresponding to the largest eigenvalues ​​of Y. To obtain matrix D', the different columns of matrix D are ordered by decreasing value, and only the columns containing the highest values ​​are retained. The number of columns retained corresponds to the new dimension, reduced compared to the original dimension. The greater the number of columns retained, the greater the proportion of the signal variance retained. Multiplication by the inverse of the matrix D' allows the inverse transformation of the dimension reduction to be performed, i.e. the reconstruction of the spectral components of the original signal from the reduced spectral components. The dimension reduction step (ivl) and the spectral component reconstruction step (iv2) are performed by matrix calculation, i.e. linear algebra.

[0167] When the dimension reduction algorithm is based on an autoencoder network, the step of reconstructing the spectral components A' 1, A'2, ..., A'p of the acquired vibration signal V is ensured by passing from the latent space to the output layer of the autoencoder. The autoencoder network ensures both the step (ivl) of dimension reduction of the spectral components as well as the step (iv2) of reconstruction of the spectral components.

[0168] [Fig.7] illustrates two examples of comparison between a sequence of spectral components of the acquired signal and a reconstructed sequence S'k of spectral components A' 1, A'2, ..., A'p. Part A corresponds to a first time acquisition window starting at a time a, part B corresponds to another time acquisition window starting at another time b. Sa and Sb designate the spectral components of the acquired signal V, in black lines. S'a and S'b designate the spectral components of the reconstructed signal, in lighter lines. For each of the two time windows, we see that the reconstructed signal is similar to the acquired signal, while being smoother, i.e. less noisy. This is the consequence of the di- reduction step mensions, which allowed the majority of the variance in the data to be retained.

[0169] The next step of the method is to compare the reconstructed vibration signal and the initially acquired vibration signal. The spectral components of the acquired signal and the spectral components reconstructed after dimension reduction are thus compared.

[0170] For this, the method comprises a step: - (iv3) Determine, for each time window Wk, a gap E between the sequence Sk of spectral components (Al, A2, ..., Ap) of the acquired vibration signal V and the reconstructed sequence S'k of spectral components (A'1, A'2, ..., A'p) of the vibration signal V, and the structural defect indicator I is based on the determined gap E.

[0171] In the absence of a structural defect, the reconstructed spectral components of the signal are close to the spectral components of the initially acquired signal. A relatively larger deviation indicates that the behavior of the mechanical system deviates from the behavior modeled in the absence of a defect, which tends to indicate the presence of a defect.

[0172] According to one embodiment of the method, the difference E between the sequence Sk of spectral components (Al, A2, ..., Ap) of the acquired vibration signal V and the reconstructed sequence S'k of spectral components (A'1, A'2, ..., A'p) of the vibration signal V is determined for all the spectral components (Al, A2, ..., Ap) of the vibration signal V.

[0173] In other words, the spectral components of the signal are reconstructed for all the spectral frequencies, and the difference between the original signal and the reconstructed signal is also determined by taking into account all the spectral frequencies, i.e. all the frequency information of the signal.

[0174] According to a variant of the method, the difference E between the sequence Sk of spectral components (Al, A2, ..., Ap) of the acquired vibration signal V and the reconstructed sequence S'k of spectral components (A'1, A'2, ..., A'p) of the vibration signal V is determined for a subset of spectral components (Ai, ..., Aj). The subset of spectral components (Ai, ..., Aj) is a subset of the sequence (S) of spectral components (Al, A2, ..., Ap) of the acquired vibration signal V, comprising strictly fewer elements than the complete set of the sequence S of spectral components.

[0175] In other words, the spectral components of the signal are also reconstructed for all spectral frequencies, but the deviation between the original signal and the reconstructed signal is in this case determined by taking into account only certain frequency ranges, which are selected. The deviation between the spectral components of the original signal and the spectral components of the reconstructed signal can thus be quantified only in the frequency ranges most favorable to the detection of a structural defect. For example, the low frequency ranges are generally more suitable for identifying a fault such as overload, or a reduction in damping efficiency. High frequency ranges are more suitable for identifying a crack in a leaf of a leaf spring.

[0176] The difference between the initially acquired signal and the reconstructed signal can be calculated using different mathematical methods.

[0177] For example, the determined gap E is the Euclidean distance between the sequence Sk of spectral components (Al, A2, ..., Ap) of the acquired vibration signal V and the reconstructed sequence S'k of spectral components (A'1, A'2, ..., A'p) of the vibration signal V.

[0178] By definition, the Euclidean distance between a vector X with components Xb ..., Xq and a vector Y with components (Yb ..., Yq) is equal to the square root of the sum, over all the components of the vectors, of the squares of the differences between the components Xk and Yk of the two vectors X and Y.

[0179] We thus have:

[0180] [Math.l]

[0181] In the next step, the detection threshold Th is determined, this threshold making it possible to differentiate a state free from structural defect and a state where a structural defect is present.

[0182] The method thus comprises a step: - (iv4) Determine a detection threshold Th, the detection threshold Th being equal to the maximum value of the determined deviation E for a subset of time windows acquired under reference conditions in which the element 10 of the running gear 20 is free from structural defects.

[0183] The learning carried out under the reference conditions makes it possible to define the detection threshold Th making it possible to establish the presence of a structural defect. The method can thus be automatically adapted to any mechanical configuration. In addition, the threshold making it possible to define a defect can be periodically updated as data acquired under reference conditions are collected. The robustness of the method is thus improved.

[0184] The presence of a defect is determined by comparison between the previously calculated deviation E and the previously determined defect threshold Th.

[0185] The method thus comprises a step: - (iv5-a) if the determined deviation E is less than or equal to the detection threshold Th, the structural defect indicator I takes a first value 10 corresponding to an absence of structural defect, - (iv5-b) if the determined deviation E is greater than the detection threshold Th, the structural defect indicator I takes a second value II corresponding to a presence of structural defect.

[0186] A significant difference between the acquired vibration signal V and the reconstructed vibration signal V' indicates the presence of a structural defect in the mechanical system. A small difference indicates that the system response is consistent with the modeled behavior in the absence of a defect.

[0187] The structural defect indicator I can take discrete values, a first value 10 corresponding to an absence of structural defect and a second value II corresponding to a presence of structural defect. More precisely, the structural defect indicator I here takes binary values.

[0188] [Fig.8] illustrates the behavior of the deviation variable E for a first vehicle with no fault and a second vehicle with a fault. More precisely, the vertical axis represents the deviation E divided by the detection threshold Th. Each point in the figure corresponds to the processing carried out for a time acquisition window. The dotted horizontal line therefore corresponds to the limit between the zone corresponding to an absence of fault, below this line, and the zone corresponding to a present fault, above the line. The first 75 measurement points correspond to operation in the reference conditions without fault. It can be noted that the point P2o is the point used for determining the fault threshold, since this point corresponds to the maximum value in the absence of fault. The value E / Th is by definition 1 for this measurement point.In the area below the dotted horizontal line, the structural defect indicator is equal to 0. The measurement points from 75 correspond to measurements made on a vehicle for which a leaf spring had a structural defect, in the form of a crack. For all these measurement points the value of the E / Th ratio is greater than 1, i.e. the deviation E is greater than the detection threshold Th. The structural defect indicator therefore takes the value 1, indicating the presence of a structural defect. The area above the dotted horizontal line corresponds to the area in which a structural defect is detected.

[0189] The primary purpose of structural defect detection is to enable the vehicle operator or manager to take corrective action. It is therefore preferable to inform the operator or manager when the proposed method detects the presence of a defect.

[0190] The method thus comprises a step: - (v) emit an alert signal in response based on the structural fault indicator I.

[0191] The alert signal emitted makes it possible to trigger an in-depth analysis of the vehicle component and to anticipate a failure.

[0192] An alert can be issued as soon as an event corresponding to a detection of structural defect is detected. It is also possible to issue an alert only if several events corresponding to a defect are detected, in order to confirm the presence of the defect.

[0193] The method can thus comprise the steps: - (vl) determine an average value Imoy of the structural defect indicator I, - (v2) emit an alert signal if a difference between the determined average value Imoy and a reference value is greater than a predetermined threshold Th2.

[0194] An isolated change in the value of the structural fault indicator does not trigger a warning signal. Repeated changes in values ​​trigger the emission of a warning signal. No warning signal is emitted when the determined average value Imoy is less than or equal to the predetermined threshold Th2. The reference value is the value corresponding to a constant absence of fault. It can be arbitrarily chosen that a state without fault corresponds to the value 0 of the indicator I. In this case, a state with fault corresponds to the value 1 of the indicator I. A different convention can be chosen, i.e. it can also be arbitrarily chosen to assign the value 1 to the indicator for operation without structural fault, and the value 0 for operation with structural fault.

[0195] The alert signal may be a fault code stored in an electronic control unit. According to a variant or additionally, the alert signal may be the illumination of a warning light. The warning light may, for example, be visible on the dashboard of the vehicle. According to another variant or also additionally, the alert signal may be a message sent to a vehicle manager.

[0196] The disclosure also relates to a device 50 for detecting a structural defect in an element 10 of a running gear 20 of a land vehicle 100, comprising: - at least one accelerometer 7, - a device 8 for acquiring the signal from the accelerometer 7, - an electronic control unit 15 configured to implement the method described above.

[0197] One or more detection devices 50 may be fitted to a given vehicle. The number of signal acquisition devices depends on the capabilities of these devices, as well as the number of accelerometers used. The number of elements to be monitored in the case of a train is naturally higher than for a van with two axles.

[0198] According to one embodiment of the detection device 50, the electronic control unit 15 is external to the vehicle 100, and the detection device 50 comprises a communication means 9 configured to transmit the acquisitions of the signal from the accelerometer 7 to the electronic control unit 15. Alternatively or additionally, the communication means 9 can be configured to transmit the spectral components of the signal from the accelerometer 7 to the electronic control unit 15.

[0199] The data acquired during the measurement of the vibration signal are thus sent to a remote server of the “cloud computing” type. The vehicle only includes the electronic equipment allowing the acquisition of the data and their transmission to the remote server(s). The data processing is carried out by computer means delocalized in relation to the vehicle. The instrumentation required by the vehicle is thus limited. The number of vehicles on which the method is implemented is not limited. The method can thus be implemented on part or all of a vehicle fleet. The alerts can for example be sent to a vehicle fleet manager. The manager can thus organize a control intervention on each vehicle for which an alert signal has been issued, and if necessary a repair intervention.Monitoring may continue until the repair intervention is carried out.

Claims

Claims

1. Method for determining an indicator (I) of a structural defect of an element (10) of a running gear (20) of a land vehicle (100), the method comprising the steps: (i) acquiring during an acquisition time window (W) a vibration signal (V) associated with the element (10) of the running gear (20), (ii) determining a sequence (S) of spectral components (Al, A2, ..., Ap) of the vibration signal (V) acquired during the acquisition time window (W), (iii) iterating steps (i) and (ii) for a set of acquisition time windows (Wi, ..., Wn) so as to obtain a set of sequences (Si, ..., Sn) of spectral components of the vibration signal (V), each sequence (Sk) of the set of sequences (Si, ..., Sn) corresponding to an acquisition time window (Wk), (iv) determining an indicator (I) of a structural defect of the element (10) of the running gear (20) from the set of sequences (Si, ..., Sn) of spectral components of the vibration signal (V) and from a detection model implementing a machine learning algorithm, in which the structural defect is a crack in the element (10) of the running gear (20), in which the element (10) of the running gear (20) is an elastic suspension element of the vehicle (100), in which the elastic suspension element (10) is a leaf spring, arranged between an axle (1) of the vehicle and a chassis (2) of the vehicle.

2. Method according to claim 1, wherein the vibration signal (V) is an acceleration signal from an acceleration sensor (7) arranged on the element (10) of the running gear (20) of the vehicle, or arranged on the axle (1), or arranged on the chassis (2) of the vehicle (100).

3. Method according to one of the preceding claims, in which the vibration signal (V) is sampled during each acquisition time window (W) at a predetermined acquisition frequency (Fa), the method comprising a step: (iül) determining the forward speed (C) of the vehicle (100), (iii2) eliminating from the set of samples of the vibration signal (V) of the acquisition time windows (Wb ..., Wn) the samples of the vibration signal (V) of the time windows corresponding to a forward speed (C) of the vehicle (100) lower than a predetermined minimum threshold (Cmin).

4. Method according to one of the preceding claims, in which step (iv) of determining an indicator (I) of structural defect comprises a step (ivl) of reducing the dimensions of the set of sequences (Sb .Sn) of spectral components (Al, A2, ..., Ap) of the vibration signal (V), so as to obtain a set of reduced sequences (SRi, ..., SRn) of combinations of spectral components of the vibration signal (V), each reduced sequence (SRk) comprising a number of spectral components less than the number p of spectral components determined in step (ii).

5. Method according to the preceding claim, in which the step (ivl) of dimension reduction of the set of sequences (Si, ..., Sn) of spectral components of the vibration signal (V) implements an unsupervised dimension reduction algorithm, the unsupervised dimension reduction algorithm generating a dimension reduction model (M).

6. Method according to the preceding claim, in which the vibration signal (V) is sampled during each acquisition time window (W) at a predetermined acquisition frequency (Fa), and in which the dimension reduction model (M) of the set of sequences (Si, ..., Sn) of spectral components of the vibration signal (V) is obtained by automatic learning of the vibration signal (V) from the samples of the vibration signal (V) of a subset of acquisition time windows acquired under reference conditions in which the element (10) of the running gear (20) is free from structural defects.

7. Method according to the preceding claim, comprising a step: (iv2) determining, for each time window (Wk), a reconstructed sequence (S'k) of spectral components (A'1, A'2, ..., A'p) of the vibration signal (V) acquired during the time window (Wk) from the reduced sequence (SRk) of combinations of spectral components and from a reconstruction model (P) obtained from the dimension reduction model (M).

8. Method according to the preceding claim, in which the reconstruction model (P) of the spectral components of the acquired vibration signal (V) is obtained by inversion of the di- reduction model (M) measurements.

9. Method according to claim 7 or 8, comprising a step: (iv3) Determining, for each time window (Wk), a deviation (E) between the sequence (Sk) of spectral components (Al, A2, ..., Ap) of the acquired vibration signal (V) and the reconstructed sequence (S'k) of spectral components (A'1, A'2, ..., A'p) of the vibration signal (V), method in which the structural defect indicator (I) is based on the determined deviation (E).

10. Method according to the preceding claim, in which the difference (E) between the sequence (Sk) of spectral components (Al, A2, ..., Ap) of the acquired vibration signal (V) and the reconstructed sequence (S'k) of spectral components (A' 1, A'2, ..., A'p) of the vibration signal (V) is determined for a subset of spectral components (Ai, ..., Aj).

11. Method according to claim 9 or 10, in which the determined deviation (E) is the Euclidean distance between the sequence (Sk) of spectral components (Al, A2, ..., Ap) of the acquired vibration signal (V) and the reconstructed sequence (S'k) of spectral components (A'1, A'2, ..., A'p) of the vibration signal (V).

12. Method according to one of claims 9 to 11, in which the vibration signal (V) is sampled during each acquisition time window (W) at a predetermined acquisition frequency (Fa), the method comprising the steps: (iv4) Determining a detection threshold (Th), the detection threshold (Th) being equal to the maximum value of the determined deviation (E) for the samples of the vibration signal (V) of a subset of acquisition time windows acquired under reference conditions in which the element (10) of the running gear (20) is free from structural defects, (iv5-a) if the determined deviation (E) is less than or equal to the detection threshold (Th), the structural defect indicator (I) takes a first value (10) corresponding to an absence of structural defects, (iv5-b) if the determined deviation (E) is greater than the detection threshold (Th),the structural defect indicator (I) takes a second value (II) corresponding to the presence of a structural defect.,

13. Method according to the preceding claim, comprising the steps: (vl) determining an average value (Imoy) of the structural defect indicator (I), (v2) emitting an alert signal if a deviation between the average value de- completed (Imoy) and a reference value is greater than a predetermined threshold (Th2).

14. Device (50) for detecting a structural defect in an element (10) of a running gear (20) of a land vehicle (100), comprising: - at least one accelerometer (7), - a device (8) for acquiring the signal from the accelerometer (7), - an electronic control unit (15) configured to implement the method according to one of the preceding claims, in which the electronic control unit (15) is external to the vehicle (100), and in which the detection device (50) comprises a communication means (9) configured to transmit the acquisitions or the spectral components of the signal from the accelerometer (7) to the electronic control unit (15).

Images