METHOD FOR MONITORING THE PHYSICAL CONDITION OF A RAIL

Description

[0001] The present invention relates to a method for monitoring the physical condition of a longitudinal element and a monitoring system for implementing this method.

[0002] Many systems or structures, such as buildings, bridges, elevators or railway networks, include longitudinal elements, particularly metallic ones, such as cables, rails, barriers or structural elements. These longitudinal elements often serve as supports or supports for elements of the system and it is therefore important to know the physical condition of these longitudinal elements, in particular their state of wear, or the variation of this physical condition over time.

[0003] Variations in the physical condition of these longitudinal elements may be due to their use itself but also to their exposure to variable climatic conditions (sunshine, rain, frost, etc.) or to the ambient environment in which they are installed, for example in the case of submerged longitudinal elements.

[0004] Railway rails, for example, wear out over time, partly due to the repeated passage of trains and partly due to their exposure to potentially extreme weather conditions. To prevent rails from breaking or cracking when trains pass over them, it is important to monitor their wear.

[0005] In the prior art, methods exist for carrying out rail monitoring. One of the methods currently used is visual inspection of the rails. Another possible method for such monitoring in document EP3509927 is based on so-called "passive" seismic, consisting of placing seismic wave sensors or other mechanical waves near the rails to identify variations in the signals propagated by the ground and which were generated by the mechanical waves linked to the contact between the wheels of the train and the rail during the passage of the train.

[0006] However, these methods only allow the detection of significant deterioration of the rail such as a crack or even a break.

[0007] It would therefore be interesting to have a solution enabling monitoring of the physical condition of any type of longitudinal element, in particular with a view to studying the variation in the physical condition of this longitudinal element over time and depending on the conditions to which this longitudinal element is or has been subjected, and to detect wear of this longitudinal element at an early stage.

[0008] Document WO 2020 / 025390 thus presents avenues for studying the condition of railway rails using guided waves in the rail. It is proposed to instrument the rail and the response signal to the guided waves received by the sensors, in particular following the passage of a train, is processed to visualize a characteristic of the condition of the rail by comparison with values ​​from theoretical or empirical models.

[0009] A first object of the invention is a method for monitoring the physical condition of a rail or other longitudinal element, the method comprising: a step of detecting mechanical waves moving along the longitudinal element by means of an array of mechanical wave sensors placed along and in contact with the longitudinal element, said array comprising at least a first pair of sensors each positioned at one end of a first portion of the longitudinal element, and a processing step comprising the determination of a plurality of simplified traces, each simplified trace resulting from the interferometry of signals delivered by the sensors of the first pair over a predetermined time period, the determination of at least a first and a second virtual trace, the first virtual trace corresponding to the sum of simplified traces determined during a first monitoring interval and the second virtual trace corresponding to the sum of simplified traces determined during a second monitoring interval,the first and second monitoring intervals being different and each comprising a plurality of predetermined time periods, and the comparison of at least the first virtual trace with the second virtual trace so as to extract at least one piece of information on the physical state of the first portion.

[0010] The monitoring method of the invention makes it possible to monitor a longitudinal element, and in particular a railway track rail, using mechanical waves moving along the latter, and in particular using trains using said rail. The sensors being placed along and in contact with the longitudinal element, they detect the mechanical waves directly, without them having passed through another material or medium such as the ground. The signal received by the sensors undergoes interferometry processing, and the interfered signal (or simplified trace) obtained is therefore solely representative of the longitudinal element and contains few, if any, parasitic signals, unlike known solutions in which sensors are placed at a distance from the longitudinal element and waves are transmitted via the ground.

[0011] In particular, in the case of a train running on a railway track and when said train is at a given distance from a point on the rail (500 m for example), it generates waves which propagate in the waveguide that constitutes the rail. This signal, very strong, is characterized by a propagation entirely located in the rail and therefore subject to carrying information on imperfections. Interferometry between the signals captured by two sensors placed on the rail during a period preceding the passage of the train (for example 30 seconds to one minute before), makes it possible to extract the propagation component of the signal, freeing itself from the component due to the source; this makes it possible to measure identified indicators such as the propagation speed for example. Then, the analysis of the temporal and spatial variations of these indicators allows the identification of a worn portion of the rail.

[0012] Furthermore, it is possible to study portions of any length depending on where the sensors are placed.

[0013] In the invention, the interfered signals are derived from the application of an interferometry technique to the signals of a pair of sensors, in particular to a pair of signals, one of the signals of this pair coming from the first sensor of a pair and the other signal of this pair coming from the second sensor of the same pair. The applied interferometry technique may be, for example, cross-correlation (sometimes called correlation), convolution, deconvolution and / or any other interferometry method. The application of an interferometry technique advantageously makes it possible to obtain usable signals because this technique makes it possible to overcome the signature of the emitted / detected signal and to extract the propagation component of said signal.

[0014] According to the method of the invention, the signals measured at each occurrence of a guided wave can be accumulated over successive periods, and the comparison can be made for each of the periods. In particular, for a railway track, the simplified traces generated by a train passing are accumulated for one day, and the evolution is evaluated by comparisons between days. This summation of signals makes it possible to reduce the mathematical processing and / or to use more robust or less precise sensors, in particular DAS technology on optical fiber. The daily or weekly, or even monthly, analysis of the modifications also facilitates “industrial” use of the invention.

[0015] A second object of the invention is a system for monitoring the physical condition of a rail or other longitudinal element, the system comprising: an array of mechanical wave sensors placed along and in contact with the longitudinal element, said array comprising at least a first pair of sensors each positioned at one end of a first portion of the longitudinal element, and a system for processing the signals from the sensors of said array of sensors, the processing system being configured to determine a plurality of simplified traces by performing interferometry of signals delivered by the sensors of the first pair of sensors over a predetermined time period, calculating at least a first and a second virtual trace, the first virtual trace corresponding to the sum of simplified traces determined during a first monitoring interval and the second virtual trace corresponding to the sum of simplified traces determined during a second monitoring interval,the first and second monitoring intervals being different and each comprising a plurality of predetermined time periods, and comparing at least the first and second virtual traces so as to extract at least one piece of information on the physical state of the first portion. Brève description des dessins

[0016] The accompanying drawings illustrate the invention: THE figures 1a et 1b represent a system for monitoring the physical state of a rail according to two embodiments of the invention; The figure 2a represents a flowchart illustrating a first monitoring method according to the invention; The figure 2b represents a first example of a signal obtained by the method of the figure 2a ; There figure 2c represents a second example of a signal obtained by the method of the figure 2a ; There figure 3a represents a flowchart illustrating a monitoring method according to a second embodiment of the invention; The figure 3b represents an example of signals obtained in the context of the method of the figure 3a ; There figure 3c represents an example of application of the second monitoring method of the invention; The figure 4a represents a flowchart illustrating a monitoring method according to a third embodiment of the invention; The figure 4b represents an example of signals obtained in the context of the method of the figure 4a ; There figure 4c represents an example of signals obtained in the context of the method of the figure 4a ; and The figure 5 represents an additional monitoring method that can be carried out using at least part of the system of the figure 1 . Description de mode(s) de réalisation

[0017] For reasons of clarity, only the elements essential for understanding the invention have been represented schematically and without respect for scale.

[0018] Likewise, in order to simplify the description, the preferred application of the invention is described, in relation to a railway rail instrumented to detect the guided waves generated by the passage of trains. However, this is a non-limiting example embodiment and the invention can be applied to any type of longitudinal element allowing propagation of mechanical waves, in particular seismic or vibratory and for which it is desired to monitor the variation in physical state. In the present invention, the mechanical waves are actively generated by a voluntary event which takes place on the longitudinal element such as for example a shock or friction, which is different from systems which carry out passive monitoring using ambient noise waves.

[0019] The methods for monitoring the physical state of a longitudinal element according to the invention and the systems making it possible to implement these methods will thus be described in the case where the longitudinal element to be monitored is a railway rail, for example, the first rail 10 shown in the figure 1a , forming with a second rail 10' a railway track 20 on which a train 30 can move.

[0020] The rail 10 may be linear or curved, the curvature being able to be in a first and / or a second direction, the first direction being horizontal and perpendicular to the longitudinal axis of the rail 10 and the second direction being vertical and perpendicular to the longitudinal axis of the rail 10.

[0021] To enable monitoring of the state of at least a first portion 10a of the rail 10, a system 12 for monitoring the physical state of the rail 10 according to the invention comprises: a network 14 of mechanical wave sensors placed along and in contact with the rail 10, the network 14 comprising at least a first pair A of sensors 1 and 2 positioned at each of the ends of the first portion 10a of the rail 10, and a system 40 for processing the signals from the sensors of said network 14 of sensors.

[0022] In the example illustrated on the figure 1a , the network 14 further comprises a second pair B of sensors formed by a third sensor 3 and a fourth sensor 4 each placed at one end of a second portion 10b of the rail 10, and a third pair C of sensors formed by the fourth sensor 4 and a fifth sensor 5 each placed at one end of a third portion 10c of the rail 10.

[0023] In the example illustrated, the sensors of the network 14 are integrated into discrete receivers placed regularly and spaced at a constant distance, for example a distance which can range from one meter to a few tens of meters. This regular distribution is advantageously chosen over the entire length of the rail which is subject to the same wear conditions. In a variant not shown, the sensors, or the receivers comprising the sensors, can be distributed irregularly, over all or part of the rail. A different distribution of the sensors, at least over certain lengths of the rail, can be interesting to take into account the fact that certain parts of the rail are likely to be subject to greater wear conditions, such as, for example, turning areas.

[0024] In the present invention, a discrete receiver comprises a sensor and a transmitting / recording unit, and is used to measure and record mechanical waves and then transmit the recorded data to the processing system 40.

[0025] Each of the sensors in the sensor network 14 may be chosen from proven technology sensors such as geophones, accelerometers and / or other mechanical wave sensors. In a preferred embodiment, all the sensors in the sensor network 14 may be identical so as to simplify processing and maintenance. According to another possible embodiment, the sensors in the sensor network 14 may be a mixture of different sensors. The sensors may be chosen, for example, based on their robustness, size, ease of deployment and required reliability: thus, for example, a network of identical microsensors (known as Micro-Electro-Mechanical Systems or MEMS) may be preferred.

[0026] In another embodiment illustrated in the figure 1b , the network 14 of sensors is produced at least partially by an optical fiber 15 secured to the rail and associated with a distributed acoustic detection device (DAS) 17 which is integrated into the processing system 40. The DAS technology which is known to those skilled in the art makes it possible to use the optical fiber as a set of equivalent sensors which would be positioned at predefined intervals along the path of the fiber.

[0027] The advantage of DAS technology is that the optical fiber 15 can extend over several tens or even hundreds of kilometers along the rail 10 or any other longitudinal element and is low cost. Furthermore, the position of each of the equivalent sensors along the optical fiber is not fixed in time and can be chosen according to the monitoring that is desired. Indeed, the detection of mechanical waves along the rail is done over the entire length of the optical fiber, and it is during the processing step that the position of each of the equivalent sensors is decided according to the portion(s) on which monitoring is desired. DAS technology associated with the use of optical fiber thus offers great flexibility of use.As with embodiments using discrete sensors, the DAS device may be configured so that the equivalent sensors are positioned regularly, irregularly, or alternately regularly or irregularly.

[0028] Any combination of sensors can be used. For example, it is possible to place an optical fiber 15 along the rail 10 on a first part of the rail and to place discrete sensors on a second part of the rail. It is also possible to place an optical fiber along the entire length of the rail or over a part of the length of the rail 10 and to position discrete sensors 6, 7 even in certain areas of the part covered by the optical fiber.

[0029] According to an embodiment illustrated below, the two sensors of the same pair of sensors are two identical accelerometers.

[0030] The sensors 1 to 5 are mechanically coupled to the rail 10, preferably by direct contact. Thus, they can directly detect mechanical, seismic or vibratory waves moving along the rail. The mechanical waves that can be detected are, for example, generated by a source which is, in this embodiment, the train 30 when it moves on the railway track 20.

[0031] Each of the sensors of the network 14 is connected to a system for processing the detected signals. In the illustrated embodiment, the processing system is represented by a processing system 40 connected (wired or wireless) to each of the sensors of the network 14 which consequently receives the signals detected by each of the sensors of the network. The processing system 40 may comprise a physical central unit, or be implemented via the internet (known as “Cloud Computing”). The processing system 40 may also comprise processing elements embedded directly on the sensors of the network 14. In fact, the processing system 40 may take any mixed configuration, with embedded processing, “cloud computing” and / or a central unit, making it possible to carry out the signal processing integrated into the invention. When the system uses DAS technology associated with an optical fiber 15 as shown in the figure 1b , the processing system 40 is associated with the distributed acoustic detection device 17.

[0032] Each sensor detects the mechanical waves over a predetermined duration and the corresponding signal is transmitted to the processing system 40. The detection can be programmed and carried out, for example, for a duration of 60s ranging from 90 seconds to 30 seconds before the trains pass the pair of sensors. Alternatively, and depending on the processing means and the nature of the sensors, the detection can be continuous (with or without sampling) and the signal fraction “interesting” for the rest of the method according to the invention is retained in the processing system 40.The processing system 40 comprises means for carrying out an interferometry step of the signals received over a predetermined time period; for example, the interferometry is carried out on a portion of the signal lasting approximately one minute shortly before the train passes, in order to take advantage of a strong and useful signal without saturating the sensors, then only one to a few seconds of correlated signal can be retained for this train passage and this, for each pair of sensors.

[0033] The system 12 makes it possible to implement three methods for monitoring the physical state of the rail 10, these three methods having in common: a step of detecting mechanical waves moving along the rail 10 by means of the network 14 of mechanical wave sensors placed along and in contact with the rail 10, said network 14 comprising at least the first pair A of sensors 1 and 2 each positioned at one end of the first portion 10a of the rail 10, and a step of processing the signals from the sensors 1 and 2 of said network 14 of sensors, said processing step comprising the determination of at least a first interfered signal resulting from the application of an interferometry technique, for example an intercorrelation, to signals delivered by the sensors 1 and 2 of the first pair A of sensors over a first predetermined time period P1.

[0034] The application of an interferometry method makes it possible to obtain usable signals, freeing oneself from the signature of the source (i.e. the train); in particular, interfered signals (or simplified traces) are obtained by applying interferometry to two signals from a pair of sensors.

[0035] To construct these simplified traces (or interfered signals), one can use cross-correlation (sometimes called correlation), convolution, deconvolution and / or any other interferometry method. In the examples to be described, the interferometry technique used is cross-correlation (or correlation).

[0036] The first and second sensors each detect a set of mechanical waves moving along the rail 10, said waves being generated by a train or any other power unit moving on the rail at least during the first time period P1. Each of the sensors 1 and 2 of the first pair A then delivers a signal corresponding to the set of mechanical waves detected and which is transmitted to the processing system 40.

[0037] The processing system 40 then generates a first intercorrelated signal resulting from a first interferometry step from the signals delivered by the sensors of the first pair A of sensors 1 and 2 over a first predetermined time period P1. This interferometry step is preferably carried out by intercorrelation, but any other method would be suitable; the aim here is to overcome the signature of the emitted / detected signal and to extract the propagation component of said signal. Thus, thanks to this step, the signal is decorrelated from the nature of the train as a source so that it will be possible to follow its evolution even if the source changes. The intercorrelation of the signals from the first sensor 1 and the second sensor 2 therefore makes it possible to obtain a first signal deconvolved from the signature of the source, this signal representing a simplified, and repeatable, trace of the propagation of mechanical waves within the rail.

[0038] The first interfered signal determined at the end of the processing step makes it possible to extract one or more pieces of information on the physical state of the portion 10a during the first time period P1. In particular, the first signal can give indications on the physical state of the portion 10a at a given time which corresponds to the first time period.

[0039] A time period according to the invention corresponds to a fraction of the total detection time (or monitoring interval) of the mechanical waves by the sensors of a pair of sensors. This time period can be more or less long depending on the desired signal quality and on the quantity and / or intensity of the mechanical waves detected by the sensors. The detection time depends on the nature of the monitoring. The duration of a time period is generally a few seconds which makes it possible to obtain a simplified trace by interferometry. On the other hand, the duration during which the sensors detect the signal can be longer and depend in particular on the duration of the monitoring interval which can be several hours, days or months. It is possible to process only a portion of the waves detected by the sensors.

[0040] A first method 100 that can be implemented using the system 12 is illustrated in the figure 2a This first method allows the monitoring of the physical state of a single portion of the rail 10 over time and will be described for the first portion 10a.

[0041] The method 100 comprises the step 110 of detecting mechanical waves moving along the rail 10, and in particular along the first portion 10a, by the first pair A of sensors 1 and 2. The method 100 also comprises the processing step 120 comprising the determination of a first simplified trace, that is to say of a first intercorrelated signal S corA-P1 resulting from a first intercorrelation of a signal S 1-P1 delivered by the first sensor 1 and of a signal S 2-P1 delivered by the second sensor 2, over the first time period ( figure 2b ).

[0042] According to method 100, the treatment step further comprises: the determination of at least one second simplified trace, i.e. a second intercorrelated signal S corA-P2 resulting from a second intercorrelation of signals S 1-P2 and S 2-P2 delivered by the sensors of the first pair A over a second predetermined time period P2 ( figure 2c ); and a comparison 123 of the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corA-P2 so as to obtain at least one piece of information on the physical state of the first portion 10a.

[0043] According to this method, the processing system 40 is configured to determine the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corA-P2 and to perform a comparison of the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corA-P2 so as to extract at least one piece of information on the physical state of the first portion 10a.

[0044] The comparison of the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corA-P2 carried out by the processing system 40 makes it possible in particular to evaluate the variation in the physical state of the first portion 10a between the first time period P1 and the second time period P2. This comparison can be done visually, or automatically by any known method, such as a correlation calculation.

[0045] In the case where the two signals are identical at the end of this comparison, the method 100 makes it possible to conclude that the physical state of the first portion 10a has not varied between the period P1 and the period P2. On the other hand, in the case where the two signals are not identical, the method 100 makes it possible to know that the physical state of the first portion 10a has varied between the first period P1 and the second period P2. It is then possible to link this signal variation with variations in certain parameters between the periods P1 and P2 such as for example a difference in climatic conditions or a long period of time elapsed which may indicate wear of the rail along this portion.

[0046] The duration and time of day (or week, month, etc.) of the time periods P1 and P2 can be chosen according to the information that one wishes to obtain on the physical state of the first portion 10a of the rail 10. If one wishes to know, for example, the variation in the physical state of the rail during a day as a function of the sunshine, the temperature, the humidity, etc., each of the periods P1 and P2 can range from several minutes to one or two hours.

[0047] According to one embodiment, the sensors 1 and 2 of the first pair A can record the mechanical waves moving along the rail continuously, or sampled at a constant frequency, during a predetermined recording phase and the time periods P1 and P2 are selected in this recording phase. In other words, the phases P1 and P2 each constitute a time fraction of the recording phase. According to this first variant associated in particular with the use of the DAS 17 technology associated with the optical fiber 15 as a network 14 of sensors, the recording phase can have a duration ranging from several seconds to several years.

[0048] According to another embodiment, the sensors 1 and 2 of the first pair A can record the mechanical waves moving along the rail only during the predetermined periods P1 and P2.

[0049] Periods P1 and P2 can be spaced apart by a duration ranging from a few seconds to a few months or, on the contrary, they can follow one another without interruption.

[0050] The first method 100 is preferably repeated so as to obtain continuous monitoring of the condition of the rail and it can therefore be repeated an unlimited number of times.

[0051] A second process 200 illustrated on the figure 3a can also be implemented using the monitoring system 12.

[0052] The second method 200 makes it possible to monitor the rail 10, and in particular to study the variation in its physical state for several durations, each comprising a plurality of time periods. This method according to the invention can, for example, make it possible to monitor a portion of rail day after day.

[0053] The second method 200 comprises the step 210 of detecting mechanical waves moving along the rail 10, said waves being generated by a train or other power unit moving on said rail by the first pair A of sensors 1 and 2. Preferably, the detection is carried out just before the train passes directly over the first portion 10a in order to avoid saturation of the pair of sensors A, for example a few tens of seconds before the train arrives on the first portion. The second method 200 also comprises the processing step 220 comprising: the determination 221 of a plurality of intercorrelated signals S corA-P1-PN corresponding to an intercorrelation of signals delivered by the sensors of the first pair A over a plurality of time periods (of identical or different durations, for example 1 second, before the passage of trains), all of the time periods forming a tracking interval of duration D. The processing system 40 thus comprises a memory for storing successive recordings of simplified traces.

[0054] The processing step also includes the sum 222 of the simplified traces, possibly normalized, in order to obtain a virtual trace S somA1-P1-PN for the monitoring interval.

[0055] The second monitoring method includes repeating this processing for other monitoring intervals, of identical or different durations, the simplified traces being obtained on identical or different criteria.

[0056] In particular, the processing step 220 comprises the determination 221' of a second plurality of intercorrelated signals S corA-P'1-P'N corresponding to a correlation of signals delivered by the sensors of the first pair A over a second plurality of time periods, as well as the sum 222' of the signals of the second plurality of intercorrelated signals S corA-P'1-P'N making it possible to obtain a second virtual trace S somA2-P'1-'PN for the second monitoring interval, of duration D' ( figure 3b ).

[0057] The processing step 220 then comprises a comparison step 223 making it possible to compare the first and second virtual traces with each other.

[0058] According to an example of application of the method 200 illustrated in the figure 3c , the method is carried out for nine monitoring intervals making it possible to obtain the nine virtual traces S som-A1 to S som-A9 which can be compared with each other to know the evolution of the state of the rail 10. In particular, the comparison of the virtual traces S som-A1 to S som-A9 makes it possible to identify that the signal varies during the monitoring intervals 1 to 9, and therefore that the physical state of the rail has been modified during these monitoring intervals. We thus note the start of wear at the level of the virtual trace S som-A7, this wear continuing thereafter. According to a possible embodiment, the method can comprise an alarm step making it possible to alert if the signal, or the difference between the signal and the “normal” average, exceeds a certain threshold.

[0059] In this exemplary embodiment, each monitoring interval lasts 1 month and the method therefore makes it possible to study the variation in the physical state of the rail 10 over 9 months. The modification of the virtual traces S som-A1 to S som-A9 continuously over the 9 months visible on the figure 3c may for example mean that the rail 10 has worn over time on the portion 10a. According to another exemplary embodiment, each monitoring interval may last 1 hour and the method may thus make it possible to study the variation in the physical state of the rail over 9 hours. In the case where the monitoring intervals run from 5 a.m. to 2 p.m. in a period of high heat, the variation in the physical state of the rail may mean that the rail has expanded under the effect of the heat. Furthermore, in the case where the next day, the signal shows the same temporal variation or continues a continuous deterioration, it is possible to consider the change as "normal" or to generate an alarm.

[0060] According to this second method 200, a time period P may for example correspond to the passage of a train on the rail 10 and each monitoring interval may be one day. Thus, at the end of a day, a first virtual trace S SomA1-P1-PN is obtained, the quality of which is improved compared to the intercorrelated signal obtained at each train passage and the method may allow daily monitoring to be carried out in the case where the method is repeated over several days, weeks, months, etc. It is noted that the virtual traces may comprise a different number of occurrences (or even simplified traces), the signals may also have been obtained over time periods of different duration.Thanks to the choice according to the invention of cumulative interferometry, it is thus possible to simplify monitoring for operators, the conditions for recording the signals being able to be chosen according to parameters inherent to the use of the infrastructure, here to the context of train circulation: the railway operator provides the train passage times (possibly the times of day or year that he considers representative), and the processing is carried out for the periods determined on this basis. Any modification to the passage schedule, whether temporary (traffic hazard) or permanent (change of season) can be taken into account directly by the system according to the invention, by adapting the time periods Pi for the determinations of simplified traces 221, 222.

[0061] According to this second method, the processing system 40 is configured to performing the sum of the intercorrelated signals for a duration comprising several time periods, and comparing the first virtual trace S SomA1-P1-PN and the second virtual trace S SomA2-P'1-P'N so as to extract at least one piece of information on the variation in the physical state of the first portion 10a between two moments (or durations).

[0062] The system and method can of course be adapted to repeat the comparisons with more than two virtual traces as shown in the figure 3c described previously.

[0063] In particular, it is possible to associate the passage of trains with the recording of the signals detected by the pairs of sensors: 90 seconds before the passage of the train, the signals detected by the pairs of sensors are transmitted for one minute to the processing system 40. The processing system 40 then performs an intercorrelation between at least a fraction of the signals of the pair of sensors and stores the result. At the end of the day, the processing system performs the sum of the recorded signals, to obtain a first virtual trace that it can normalize, using a spectral whitening step for example. The daily virtual traces are then compared in order to evaluate a change in the physical state of the portion of rail.

[0064] This process is particularly suitable for fiber optic sensors associated with a DAS device that records continuously, the summation amplifying the signal. Furthermore, depending on the changes observed between the virtual traces, it is possible to redefine the position of the equivalent sensors for better precision.

[0065] A third process 300 illustrated on the figure 4a can also be implemented using the 12 monitoring system.

[0066] The third method 300 comprises the detection step 310 which comprises the detection 310 A of the mechanical waves moving along the rail 10, and in particular along the first portion 10a, by the first pair A of sensors 1 and 2. The method 300 also comprises the processing step 320 comprising the determination of a first intercorrelated signal S corA-P1 resulting from a first intercorrelation of a signal S 1-P1 delivered by the first sensor 1 and a signal S 2-P1 delivered by the second sensor 2, over the first time period ( figure 4b ).

[0067] According to the third method 300, the detection step 310 further comprises the detection 310 B of the mechanical waves by means of the second pair B of sensors 3 and 4 of the network 14. Furthermore, the processing step further comprises: the determination 320 B of at least one second intercorrelated signal S corB-P1 resulting from a second correlation of signals S 3-P1 and S 4-P1 delivered by the sensors of the second pair B of sensors over the first time period P1 ( figure 4c ), and a comparison 330 of the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corB-P1 so as to extract at least one piece of information on the physical state of the first portion 10a and / or the second portion 10b.

[0068] According to this third method, the processing system 40 is configured to determining at least the second intercorrelated signal S corB-P1, and comparing the first intercorrelated signal S corA-P1 and the second intercorrelated signal S corB-P1 so as to extract at least one piece of information on the physical state of the first portion 10a and / or the second portion 10b.

[0069] In the case where the first and second intercorrelated signals S corA-P1 and S corB-P1 are different, it is possible to conclude that one of the two portions is more worn or is more damaged than the other by the conditions, for example climatic or usage, to which the two portions are subjected. The comparison can be made in a similar way to that previously described. On the other hand, in the case where the two signals are identical, it is possible to conclude that the two portions present the same state of wear and / or the same evolution following exposure to particular conditions of temperature, humidity, etc.

[0070] This third method has been described in the case where the first and second portions are spaced apart as is the case for portions 10a and 10b. However, this method 200 can also be applied to a system in which the first and second portions are contiguous and in which one of the sensors in the sensor array is common to the first and second pairs. As illustrated in the figure 1 , the first portion can be portion 10b and the second portion can be portion 10c. In this case, the first and second portions have in common the fourth sensor 4 which constitutes the second sensor of pair B and the first sensor of pair C.

[0071] The method 300 is not limited to two rail portions and can instead be applied to a greater or lesser number of portions. According to an exemplary embodiment, the method 300 can be applied to the system 12 illustrated in the figure 1 and which comprises the first portion 10a, the second portion 10b and a third portion 10c.

[0072] According to this exemplary embodiment, the three portions 10a, 10b and 10c are identical. Implementing the method 300 on these three portions can make it possible to obtain a signal for each of the portions, the three signals then being able to be compared. For example, obtaining two identical signals and a third signal different from the other two can indicate the presence of at least one anomaly (wear, breakage, deformation, etc.) on one of the three portions. In the case where three different signals are obtained, the three portions or two of the three portions may have one or more anomalies.

[0073] According to one embodiment, it is possible to carry out the daily monitoring described previously (according to the second method) so as to carry out monitoring of the physical state of each of the portions over the days and also to see if the physical state of one or more portions varies differently from the other portions over the days. In the different embodiments that have been described, the sensors were placed only along the rail 10 of the track 20. However, the sensors can be placed on both rails 10 and 10' of the track 20, according to the same distribution or according to a different distribution.

[0074] Additionally to the embodiments previously described, the waves emitted by the train 30 passing on the track 20 can be used to carry out monitoring of the subsoil between two rails 10' and 10" of a track 22 which extends parallel to the track 20 according to the method described in the patent application WO2020 / 021177. For this, pairs of sensors 1"-1‴ to 5"-5"' can be placed on rails 10" and 10"' of the track 22 ( figure 5 ). When train 30 passes over track 20 and in the absence of a train passing over track 22, the pairs of sensors placed on track 22 allow the monitoring of the subsoil between the 10" rail and the 10"' rail by interferometry. In the same way, the monitoring of the subsoil between the two rails 10 and 10' of track 20 could be carried out by placing sensors on the 10' rail and using the waves emitted by a train passing over track 22.

[0075] This combination made possible by the invention is particularly advantageous from an economic point of view, the profitability of railway instrumentation being thus increased. In addition, the geophysical imaging of the subsoil thus produced is as close as possible to the rails and allows focusing under the track itself. This constitutes a major advantage for increasing the resolution and positioning of the image.

[0076] Furthermore, in addition to the embodiments previously described, the sensors 1 to 5 of the rail 10 (or in general the sensors placed on a rail and allowing the implementation of the methods of the invention) can also be used to carry out a modal analysis of the rail along which they are placed. The modal analysis method is known to those skilled in the art and in particular allows the study of the dynamic model of the rail along which the sensors are placed.

[0077] Additionally to the previously described embodiments and when the sensor network is formed by an optical fiber, the optical fiber can be used to measure the deformation of the longitudinal element against which it is placed, for example a railway rail. For this, the optical fiber is associated with a DAS device.

[0078] According to a possible exemplary embodiment, the method 300 can be carried out for a certain number of initial portions, for example with a length of 100 meters. When an anomaly is identified on one of these initial portions, a more detailed analysis can be carried out by dividing this initial portion into several secondary portions, for example with a length of 10 meters in order to identify the secondary portion(s) responsible for the anomaly of the initial portion. The initial instrumentation of the track can comprise “dormant” sensor systems initially; advantageously, it is the programming of the DAS device which is modified to allow the more detailed segmentation. If more detailed continuous monitoring is envisaged for a portion, it is also possible to add sensors later, within a network of discrete sensors or in combination with an installed optical fiber.

[0079] According to a possible embodiment, the first method 100 and the second method 200 can be combined. For this, it is possible to carry out the monitoring method for several time periods, and this, for several portions. It is then possible to compare the signals of several portions and this for several time periods which can make it possible to study the variation of physical state of each portion over time and also to compare the portions with each other in order to detect possible anomalies. Similarly, it is possible to combine the second and third methods in order to monitor several portions for different monitoring intervals.

[0080] The methods 100, 200 and 300 of the invention, whether taken alone or in combination, thus make it possible to carry out fine measurements of slight variations in the physical state of a rail and therefore to provide a regular diagnosis, for example daily, of the physical state of a rail. Advantageously, the method of correlation of the signals by interferometry makes it possible to extract the propagation component of the signal by eliminating the component due to the source. It is then possible to measure identified indicators of the signal such as the propagation speed, the amplitude, the shape of the wave, the resonance frequency, etc., for each portion monitored. The analysis of the variations of these indicators over several time periods for a given portion of the rail and / or for different portions of the rail makes it possible to identify a worn portion of the rail.

[0081] In the embodiment just described, the longitudinal element is a rail and the train or trains passing over the rail are used as a source of mechanical waves. However, the longitudinal element may be any type of longitudinal element associated with a source of mechanical waves such as, for example, cables of a suspension bridge or a structural element of a bridge for which the source of mechanical waves may be the passage of cars over the bridge, or an elevator cable for which the source of mechanical waves may be the movement of the elevator.

[0082] According to other embodiments, the longitudinal element may not be associated with a source of mechanical waves by its use and a source of mechanical waves may then be added to implement the method according to the invention. The source may then be any type of device causing shocks and / or vibrations on the longitudinal element such as for example a device which would tap on the longitudinal element at a regular or irregular frequency.

Claims

1. Method (200) for monitoring the physical state of a longitudinal element (10), characterized in that it comprises: - a step (210) of detection of the mechanical waves moving along the longitudinal element (10) by means of an array (14) of mechanical wave sensors placed along and in contact with the longitudinal element, said array (14) comprising at least a first pair (A) of sensors (1, 2) each positioned at one end of a first portion (10a) of the longitudinal element (10), and - a processing step (220) comprising • determining (221) a plurality of simplified traces (ScorAP1-Pi), each simplified trace (ScorAPi) resulting from the interferometry of signals (S1-P1, S2-P1) delivered by the sensors of the first pair A over a predetermined time period (Pi), the method being characterized in that the processing step comprises: • determining (222) at least a first virtual trace and a second virtual trace (SsomA1, SsomA2), the first virtual trace (SsomA1) corresponding to the sum of simplified traces determined during a first monitoring interval and the second virtual trace (SsomA2) corresponding to the sum of simplified traces determined during a second monitoring interval, the first and the second monitoring intervals being different and each comprising a plurality of predetermined time periods, and • comparing (223) at least the first virtual trace (SsomA1) with the second virtual trace (SsomA2) in such a way as to extract at least one piece of information on the physical state of the first portion (10a).

2. Method according to Claim 1, characterized in that the interferometry consists of cross-correlation.

3. Method according to Claim 1 or 2, characterized in that the array (14) of sensors comprises an optical fibre (15) associated with a distributed acoustic sensing (DAS) device (17), said first pair of sensors (1, 2) being defined on the optical fibre by the DAS system during the processing step (220).

4. Method according to any one of the preceding claims, characterized in that the determination of the virtual traces comprises a normalization step.

5. Method according to Claim 4, characterized in that the normalization step is carried out by spectral whitening.

6. Method according to any one of the preceding claims, characterized in that the detection of the mechanical waves is carried out continuously.

7. Method according to any one of the preceding claims, characterized in that it is carried out for a plurality of pairs of sensors.

8. Method according to any one of the preceding claims, characterized in that the monitoring intervals are of identical duration, in particular equal to one day.

9. Method according to any one of the preceding claims, characterized in that the longitudinal element (10) is a rail of a railroad track, the mechanical waves detected corresponding to the passage of a train over said rail.

10. Method according to Claim 9, characterized in that each time period corresponds to the passage of a train over the rail, detection being carried out before the train arrives at the first portion.

11. System (12) for monitoring the physical state of a longitudinal element (10), characterized in that it comprises: - an array (12) of mechanical wave sensors placed along and in contact with the longitudinal element, said array (12) comprising at least a first pair (A) of sensors each positioned at one end of a first portion (10a) of the longitudinal element (10), and - a system (40) for processing the signals from the sensors of said array (12) of sensors, the processing system (40) being configured to • determine a plurality of simplified traces (ScorAP1-Pi) by interferometry of signals (S1-P1, S2-P1) delivered by the sensors (1, 2) of the first pair (A) of sensors over a predetermined time period (Pi), characterized in that the processing system is configured to • calculate at least a first virtual trace and a second virtual trace (SsomA1, SsomA2), the first virtual trace (SsomA1) corresponding to the sum of simplified traces determined during a first monitoring interval and the second virtual trace (SsomA2) corresponding to the sum of simplified traces determined during a second monitoring interval, the first and the second monitoring intervals being different and each comprising a plurality of predetermined time periods, and • compare at least the first and the second virtual traces in such a way as to extract at least one piece of information on the physical state of the first portion (10a) .

12. System according to Claim 11, characterized in that at least one sensor of the array (12) of sensors is a geophone or an accelerometer.

13. System according to either of Claims 11 or 12, characterized in that the array (12) of sensors is produced at least partially using an optical fibre associated with a distributed acoustic sensing (DAS) device.

14. System according to any one of Claims 11 to 13, characterized in that the system comprises a source of mechanical waves configured to generate the mechanical waves detected by the array (12) of sensors.

15. System according to any one of Claims 11 to 14, characterized in that the longitudinal element (10) is a rail of a railroad track, and in that the array (12) of sensors is capable of detecting the mechanical waves generated by the passage of at least one train over the track.