METHOD AND DEVICE FOR DETECTING ANOMALIES ALONG A RAILWAY TRACK

A method for detecting anomalies along a railway track and a device for detecting anomalies along a railway track are provided .

Fiber optic sensors can be employed in railway monitoring . For this purpose , a laser pulse is fed into an optical fiber extending along a railway track . By analyzing the backscattered signal noise on and around the railway track can be detected . Noise can relate to anomalies along the railway track . The backscattered signal can be employed to determine di f ferent parameters of the railway track or of the movement of the rail vehicles . For example the velocity or the position of the rail vehicles or defects of the railway track can be determined .

However, parameters determined from backscattered signals might have an error rate . This leads to inaccuracies in the monitoring .

It is an obj ective to provide a method for detecting anomalies along a railway track with an improved accuracy . It is further an obj ective to provide a device for detecting anomalies along a railway track with an improved accuracy .

These obj ectives are achieved with the independent claims . Further embodiments are the subj ect of dependent claims . According to at least one embodiment of the method for detecting anomalies along a railway track, the method comprises detecting at least one first sensor signal by a fiber optic sensor for a measurement segment of the fiber optic sensor, the fiber optic sensor being arranged along the railway track . The first sensor signal can comprise a backscattered signal or backscattered signals of an input signal which is provided to the fiber optic sensor . The fiber optic sensor can comprise an optical fiber . The optical fiber can be arranged within the ground close to the railway track . It is further possible that the optical fiber is arranged above the ground close to the railway track . The optical fiber can extend approximately parallel to the railway track . The input signal can be an optical signal , for example a laser pulse . The input signal can be provided to the optical fiber at an input of the optical fiber . A small part of the laser light is reflected back to the input since the laser light is scattered at scatter sites , as for example impurities in the optical fiber which can be natural or arti ficial . Changes in the backscattered signal are related to physical changes in the optical fiber which can be caused by noise , structure-borne noise , vibrations or soundwaves along the optical fiber . Therefore , a backscattered signal can be detected when a rail vehicle is moving on the railway track . By evaluating the backscattered signal , the location of the noise or the rail vehicle along the optical fiber can be determined .

The length of the fiber optic sensor can amount to several kilometers or several hundreds of kilometers . The fiber optic sensor can be divided into a plurality of measurement segments . The measurement segment can be one of the plurality of measurement segments . Each measurement segment corresponds to a predefined length along the fiber optic sensor . This means , each measurement segment directly adj oins another measurement segment . The measurement segments can all have the same length . For example the measurement segments each have a length of a few meters , for example less than 10 m .

The first sensor signal can comprise backscattered signals from the measurement segment . This can mean that the first sensor signal comprises backscattered signals originating from the measurement segment . In other words , the first sensor signal comprises backscattered signals that were scattered at the position of the measurement segment .

The method further comprises detecting at least one second sensor signal by the fiber optic sensor for the measurement segment after detecting the first sensor signal . The second sensor signal can comprise a backscattered signal or backscattered signals of an input signal which is provided to the fiber optic sensor . The second sensor signal can comprise backscattered signals from the measurement segment . This can mean that the second sensor signal comprises backscattered signals originating from the measurement segment . In other words , the second sensor signal comprises backscattered signals that were scattered at the position of the measurement segment . The second sensor signal can be detected at a later point in time than the first sensor signal .

The method further comprises determining a first di f ference signal where the first di f ference signal relates to the di f ference between an average sensor signal and the first sensor signal , wherein the average sensor signal relates to an average of previous sensor signals detected by the fiber optic sensor for the measurement segment before detecting the first sensor signal . The average sensor signal can be stored on a memory device or in a database . The previous sensor signals can be detected in the same way as the first sensor signal . The previous sensor signals can all comprise backscattered signals originating from the measurement segment . The average sensor signal can relate to the average of at least ten sensor signals detected by the fiber optic sensor for that measurement segment for which the first sensor signal is detected . The average sensor signal can relate to the average of the amplitude of the previous sensor signals . The previous sensor signals can be detected during a calibration phase . The first di f ference signal can relate to the di f ference between the amplitude of the average sensor signal and the amplitude of the first sensor signal .

The method further comprises determining a second di f ference signal where the second di f ference signal relates to the di f ference between the average sensor signal and the second sensor signal . The second di f ference signal can relate to the di f ference between the amplitude of average sensor signal and the amplitude of the second sensor signal .

The method further comprises providing an alarm signal for the case that a confidence condition is ful filled, wherein the confidence condition requires at least that the first di f ference signal and the second di f ference signal are each larger than a predefined threshold signal or the confidence condition requires at least that the absolute value of the first di f ference signal and the absolute value of the second di f ference signal are each larger than a predefined threshold signal . The first sensor signal can be smaller or larger than the average sensor signal . The second sensor signal can be smaller or larger than the average sensor signal . The alarm signal can be provided to a person or to a railway monitoring or coordination division . The alarm signal can comprise the information that the confidence condition is ful filled . The alarm signal can further comprise the information where the measurement segment is located along the optical fiber . The alarm signal can comprise the information which position along the railway track is the closest to the measurement segment . The threshold signal can be a measure for a situation deviating from normal operation of railway traf fic .

The predefined threshold signal can be determined by an optimi zation program . This can mean, that an algorithm can be employed to determine the predefined threshold signal . The value of the predefined threshold signal can be determined in such a way that i f a di f ference signal is larger than the threshold signal , an anomaly took place along the railway track with a certain probability . It is also possible that the threshold signal is determined manually .

Fiber optic sensing can be employed to monitor di f ferent conditions of the railway track or to monitor railway traf fic . By analyzing the backscattered signals detected by fiber optic sensing di f ferent situations occurring at the measurement segment or in the vicinity of the measurement segment can be di f ferentiated . However, di f ferent impacts can lead to errors in determining which situation took place during the detection of the backscattered signals . In order to improve the accuracy of determining which situation took place during the detection of the backscattered signals , the confidence condition is introduced in the method described herein . A particular situation to be determined can be identi fied from the di f ference between the average sensor signal and the first sensor signal or the second sensor signal . The average sensor signal gives the value of backscattered signals detected under normal circumstances , this means for example without any defects or undesired events . I f the first sensor signal deviates by more than the predefined threshold value from the average sensor signal , it can be assumed that an unusual or undesired event , this can mean an anomaly, took place during the detection of the first sensor signal . I f the second sensor signal deviates by more than the predefined threshold value from the average sensor signal , it can be assumed that an unusual or undesired event , this can mean an anomaly, took place during the detection of the second sensor signal . However, there is a certain probability for each of the first sensor signal and the second sensor signal that there is a false alarm . This can mean, that for each of the first sensor signal and the second sensor signal it is possible that they deviate by more than the predefined threshold value from the average sensor signal but no anomaly took place .

In order to increase the accuracy of determining i f an anomaly took place , it is assumed that only i f both the first di f ference value and the second di f ference value are larger than the predefined threshold value , an anomaly took place . Therefore , by introducing the confidence condition, the accuracy of detecting anomalies along the railway track is increased . The alarm signal is only provided under the condition that confidence is built up by the confidence condition being ful filled . Alternatively to providing an alarm signal, a visual indication of an average value of the first difference signal and the second difference signal is provided.

Average value can mean an arithmetic average value, also referred to as mean value, a geometric mean value, a median or other ways of calculating an average.

In one embodiment, the visual indication is based on how many standard deviations the average value of the difference values is away from the average sensor signal. This can be visualized for example by different colors, different y-axis values of lines or curves, or other visual indications that are easily comprehensible, e.g. different hatching or different line types.

According to at least one embodiment of the method an anomaly can be at least one of the following: a defect of the railway track at the position that is the closest to the measurement segment, a change of the condition of the railway track at the position that is the closest to the measurement segment, mechanical vibrations at or around the position of the railway track that is the closest to the measurement segment. That there is a defect of the railway track can mean that there is a defect of the rail of the railway track. If the railway track has a defect, the backscattered signal detected during the passage of a rail vehicle over the defect can be different from the average sensor signal for the same position. For example, if a defect is present, the amplitude of the backscattered signal can be increased. This can lead to a difference signal being larger than the threshold signal. In this way, the anomaly can be identified. A change in the condition of the railway track can be a change of the condition of the rail or of other components of the railway track . The condition of the rail can be a condition of wear of the rail . For di f ferent conditions of the railway track, the backscattered signal can deviate from the average sensor signal . In this way, the anomaly can be identi fied .

Mechanical vibrations at or around the railway track can be caused by at least one of movements of vehicles , footsteps of persons , manual or machine digging, working parties , movement of animals or environmental events such as rock falls or landslides . Other examples are theft or vandalism . I f mechanical vibrations occur at or around the railway track, the backscattered signal is di f ferent from the backscattered signal of the situation where no mechanical vibrations occur . In this way, the anomaly can be identi fied .

For all these di f ferent types of anomalies it might be desired to detect i f they occurred . For example , in case of a defect at the rail , the passage of rail vehicles at this position can only be allowed after the rail is repaired . This increases the safety in railway traf fic . As another example , i f a rock fall is detected, rail vehicles that are supposed to pass the position of the rock fall can be warned or stopped from passing this position .

According to at least one embodiment of the method the first sensor signal and the second sensor signal are detected during the passage of a rail vehicle over the position of the railway track that is the closest to the measurement segment . This can mean, that the first sensor signal is detected during the passage of a part of the rail vehicle that is di f ferent from another part of the rail vehicle , wherein the second sensor signal is detected during the passage of the other part of the rail vehicle . The passage of the rail vehicle causes mechanical vibrations . The mechanical vibrations change the backscattered signal detected by the fiber optic sensor . Detecting the first sensor signal and the second sensor signal during the passage of a rail vehicle has the advantage that defects and/or changes of the railway track, in particular of the rail , can be detected . From the average sensor signal it is known how the backscattered signal looks like for the measurement segment . I f there is a defect or a change of the railway track, the first sensor signal and the second sensor signal will deviate from the average sensor signal . I f the first sensor signal and the second sensor signal deviate by more than the threshold signal from the average sensor signal , a defect or change of the railway track is regarded as severe enough so that it should be noted in order to avoid accidents . Without the rail vehicle passing over the position of the defect or change , the defect or change might not be recogni zed, since the defect or change itsel f in most cases do not change the backscattered signal . An advantage of detecting the first sensor signal and the second sensor signal during the passage of only one rail vehicle is , that the confidence condition can be ful filled after the passage of only one rail vehicle . It is thus not necessarily required to wait for another rail vehicle to pass the position that is the closest to the measurement segment .

According to at least one embodiment of the method the first sensor signal is detected during the passage of a rail vehicle over the position of the railway track that is the closest to the measurement segment and the second sensor signal is detected during the passage of a further rail vehicle over the position of the railway track that is the closest to the measurement segment . This can mean, that two rail vehicles are employed to ful fill the confidence condition . With this , the influence of the particular rail vehicle on the backscattered signals from which it is determined i f the confidence condition is ful filled, is reduced .

According to at least one embodiment of the method the alarm signal indicates that an anomaly is detected along the measurement segment . This can mean, that the alarm signal indicates that an anomaly is detected along the measurement segment with a particular certainty . The information that an anomaly is detected is thus advantageously provided . It is therefore possible to take necessary action in order to avoid accidents .

In another embodiment , instead of providing an alarm signal , a visual indication is provided, e . g . as a 1-D diagram or 2-D diagram, e . g . on a display . The visual indication may indicate an anomaly along the measurement segment .

According to at least one embodiment of the method the steps of the method are carried out for a plurality of di f ferent measurement segments along the fiber optic sensor . The steps of the method can be carried out for the plurality of di f ferent measurement segments in the same way as for the measurement segment . This enables to detect anomalies along the whole railway track .

According to at least one embodiment of the method the confidence condition further requires that the first di f ference signal and the second di f ference signal are each larger than the predefined threshold signal for the first sensor signal and the second sensor signal being detected within a predefined time frame . This can mean that for determining i f an anomaly occurred, the principle of erosion in time , in particular time based erosion in time , can be employed . Thus , for the confidence condition to be ful filled, the first sensor signal and the second sensor signal have to be detected within a predefined time frame . The length of the predefined time frame can depend on the frequency of expected anomalies to be detected . It is also possible that the length of the predefined time frame depends on the frequency of rail vehicles passing the measurement segment . With the first sensor signal and the second sensor signal being required to be detected within the predefined time frame , the accuracy of determining i f an anomaly occurred can be increased . Only i f the first sensor signal and the second sensor signal are detected in a particular temporal proximity, the confidence condition is regarded to be ful filled . With this , backscattered signals that deviate from the average sensor signal and that are detected at considerably di f ferent points in time , do not ful fill the confidence condition since these backscattered signals might have been detected in two di f ferent situations that do not relate to the same anomaly .

According to at least one embodiment of the method the confidence condition further requires that between detecting the first sensor signal and detecting the second sensor signal less than three third sensor signals are detected by the fiber optic sensor for the measurement segment , wherein a third di f ference signal , which relates to the di f ference between the average sensor signal and the third sensor signal , respectively, is smaller than the threshold signal for each third sensor signal . The third sensor signal can be detected in the same way as the first sensor signal and the second sensor signal . The confidence condition can further require that between detecting the first sensor signal and detecting the second sensor signal the less than three third sensor signals are the only signals detected . This can mean, that between the first sensor signal and the second sensor signal no other signals except for less than three third sensor signals are detected . In this way, it is guaranteed that the detection of the first sensor signal and the detection of the second sensor signal are not spaced too far from each other in time which increases the probability that they relate to the same anomaly .

According to at least one embodiment of the method the confidence condition further requires that between detecting the first sensor signal and detecting the second sensor signal less than five or less than ten third sensor signals are detected by the fiber optic sensor for the measurement segment , wherein a third di f ference signal , which relates to the di f ference between the average sensor signal and the third sensor signal , respectively, is smaller than the threshold signal for each third sensor signal .

According to at least one embodiment of the method the confidence condition further requires that during detection of the first sensor signal a part of a rail vehicle passes over the position of the railway track that is the closest to the measurement segment and that during detection of the second sensor signal another part of the same rail vehicle passes over the position of the railway track that is the closest to the measurement segment . In this way, advantageously it is possible to fulfill the confidence condition with only one rail vehicle.

According to at least one embodiment of the method a counter is incremented once for each difference signal being larger than the threshold signal and decremented once for each difference signal being smaller than the threshold signal. The difference signal can be a first difference signal or a second difference signal. For determining if an anomaly occurred along the railway track the principle of erosion in time, in particular counter based erosion in time, can be employed. By incrementing the counter for each difference signal being larger than the threshold signal once and by decrementing the counter for each difference signal being smaller than the threshold signal once, the situations where the difference signal is larger than the threshold signal and where the difference signal is smaller than the threshold signal are equally weighted. In other words, a symmetric counter can be employed. By employing the counter, confidence is built up before the alarm signal is provided.

According to at least one embodiment of the method a counter is incremented at least once for each difference signal being larger than the threshold signal and decremented at least once for each difference signal being smaller than the threshold signal.

According to at least one embodiment of the method a counter is incremented twice for each difference signal being larger than the threshold signal and decremented once for each difference signal being smaller than the threshold signal. By incrementing the counter for each difference signal being larger than the threshold signal twice and by decrementing the counter for each difference signal being smaller than the threshold signal once, the situations where the difference signal is larger than the threshold signal and where the difference signal is smaller than the threshold signal are weighted differently. In other words, an asymmetric counter can be employed. By employing the counter, confidence is built up before the alarm signal is provided.

According to at least one embodiment of the method the confidence condition further requires that the counter reaches three counts. Thus, the alarm signal is only provided if the difference signal is larger than the threshold signal at least twice or at least three times. With this, confidence is built up before the alarm signal is provided. This increases the accuracy of detecting anomalies.

According to at least one embodiment of the method the confidence condition further requires that the counter reaches at least three counts or more than three counts or more than ten counts.

According to at least one embodiment of the method a counter is incremented at least one more time for each difference signal being larger than the threshold signal than for each difference signal being smaller than the threshold signal. This can mean, that the number by which the counter is incremented for each difference signal being larger than the threshold signal is larger by least one than the number by which the counter is incremented for the difference signal being smaller than the threshold signal. It is also possible that the counter is incremented at least one more time for each difference signal being larger than the threshold signal than the counter is decremented for each difference signal being smaller than the threshold signal . Thus , the counter can be incremented by any number for each di f ference signal being larger than the threshold signal . The counter can be decremented by any number for each di f ference signal being smaller than the threshold signal . By employing the counter, confidence can be built up before the alarm signal is provided .

According to at least one embodiment of the method the threshold signal is larger than the variance or standard deviation of the average sensor signal . The threshold signal can therefore be a measure for how much the backscattered signals usually deviate from their average value . The threshold signal can then be slightly larger than the variance or the standard deviation of the average sensor signal . In this way, unusual deviations from the average sensor signal are detected . This enables to detect anomalies along the railway track .

Furthermore , a device for detecting anomalies along a railway track is provided . The device for detecting anomalies along a railway track can preferably be employed in the method described herein . This means all features disclosed for the method for detecting anomalies along a railway track are also disclosed for the device for detecting anomalies along a railway track and vice-versa .

According to at least one embodiment of the device for detecting anomalies along a railway track, the device comprises an evaluation unit that is connectable to a fiber optic sensor being arranged along the railway track . This can mean, that the evaluation unit is configured to be connected to the fiber optic sensor . The evaluation unit comprises a detection unit that is configured to receive at least one first sensor signal detected by the fiber optic sensor for a measurement segment of the fiber optic sensor and to receive at least one second sensor signal detected by the fiber optic sensor for the measurement segment after detecting the first sensor signal . The evaluation unit can comprise an input for receiving the first sensor signal and the second sensor signal .

The evaluation unit comprises a subtraction unit that is configured to determine a first di f ference signal where the first di f ference signal relates to the di f ference between an average sensor signal and the first sensor signal , and to determine a second di f ference signal where the second di f ference signal relates to the di f ference between the average sensor signal and the second sensor signal , wherein the average sensor signal relates to an average of previous sensor signals detected by the fiber optic sensor for the measurement segment before detecting the first sensor signal .

The evaluation unit comprises an alarm unit that is configured to provide an alarm signal for the case that a confidence condition is ful filled, wherein the confidence condition requires at least that the first di f ference signal and the second di f ference signal are each larger than a predefined threshold signal .

As the device for detecting anomalies along a railway track can be employed in the method described herein, the device for detecting anomalies along a railway track has the same advantages as the method for detecting anomalies along a railway track . According to at least one embodiment of the device for detecting anomalies along a railway track, the device comprises a counter that is configured to be incremented once or twice for each di f ference signal being larger than the threshold signal and decremented once for each di f ference signal being smaller than the threshold signal . The counter can be connected with the alarm unit . Thus , an alarm signal can be provided i f the confidence condition is ful filled .

The following description of figures may further illustrate and explain exemplary embodiments . Components that are functionally identical or have an identical ef fect are denoted by identical references . Identical or ef fectively identical components might be described only with respect to the figures where they occur first . Their description is not necessarily repeated in successive figures .

With figure 1 an exemplary embodiment of the method for detecting anomalies along a railway track is described .

Figure 2 shows an exemplary embodiment of a device for detecting anomalies along a railway track .

Figure 3 shows a fiber optic sensor .

With figures 4 and 5 further exemplary embodiments of the method for detecting anomalies along a railway track are described .

Figure 6 shows the principle of fiber optic sensing . With figure 7 it is shown how the principle of fiber optic sensing is employed in the method for detecting anomalies along a railway track 20 .

Figures 8 and 9 show exemplary embodiments of providing visual indications .

With figure 1 an exemplary embodiment of the method for detecting anomalies along a railway track 20 is described . The method comprises detecting at least one first sensor signal F by a fiber optic sensor 21 for a measurement segment 22 of the fiber optic sensor 21 in a first step S I . The fiber optic sensor 21 is arranged along the railway track 20 .

In a second step S2 , at least one second sensor signal S is detected by the fiber optic sensor 21 for the measurement segment 22 after detecting the first sensor signal F . The first sensor signal F and the second sensor signal S can be detected during the passage of a rail vehicle over the position of the railway track 20 that is the closest to the measurement segment 22 . This can mean that , the confidence condition further requires that during detection of the first sensor signal F a part of a rail vehicle passes over the position of the railway track 20 that is the closest to the measurement segment 22 and that during detection of the second sensor signal S another part of the same rail vehicle 32 passes over the position of the railway track 20 that is the closest to the measurement segment 22 . Alternatively, the first sensor signal F can be detected during the passage of a rail vehicle over the position of the railway track 20 that is the closest to the measurement segment 22 and the second sensor signal S can be detected during the passage of a further rail vehicle over the position of the railway track

20 that is the closest to the measurement segment 22.

In a third step S3, a first difference signal FD is determined where the first difference signal FD relates to the difference between an average sensor signal AS and the first sensor signal F, wherein the average sensor signal AS relates to an average of previous sensor signals detected by the fiber optic sensor 21 for the measurement segment 22 before detecting the first sensor signal F.

In a fourth step S4, a second difference signal SD is determined where the second difference signal SD relates to the difference between the average sensor signal AS and the second sensor signal S.

In a fifth step S5, an alarm signal AL is provided for the case that a confidence condition is fulfilled, wherein the confidence condition requires at least that the first difference signal FD and the second difference signal SD are each larger than a predefined threshold signal or the confidence condition requires at least that the absolute value of the first difference signal FD and the absolute value of the second difference signal SD are each larger than a predefined threshold signal. The alarm signal AL indicates that an anomaly is detected along the measurement segment 22.

An anomaly can be at least one of the following: a defect of the railway track 20 at the position that is the closest to the measurement segment 22, a change of the condition of the railway track 20 at the position that is the closest to the measurement segment 22, mechanical vibrations at or around the position of the railway track 20 that is the closest to the measurement segment 22 .

The steps of the method can be carried out for a plurality of di f ferent measurement segments 22 along the fiber optic sensor 21 .

The confidence condition can further require that the first di f ference signal FD and the second di f ference signal SD are each larger than the predefined threshold signal for the first sensor signal F and the second sensor signal S being detected within a predefined time frame . The threshold signal can be larger than the variance or standard deviation of the average sensor signal AS .

Figure 2 shows an exemplary embodiment of the device 24 for detecting anomalies along a railway track 20 . The device 24 comprises an evaluation unit 25 that is connectable to a fiber optic sensor 21 being arranged along the railway track 20 . The evaluation unit 25 comprises a detection unit 26 that is configured to receive at least one first sensor signal F detected by the fiber optic sensor 21 for a measurement segment 22 of the fiber optic sensor 21 and to receive at least one second sensor signal S detected by the fiber optic sensor 21 for the measurement segment 22 after detecting the first sensor signal F . The evaluation unit 25 comprises a subtraction unit 27 that is configured to determine a first di f ference signal FD where the first di f ference signal FD relates to the di f ference between an average sensor signal AS and the first sensor signal F, and to determine a second di f ference signal SD where the second di f ference signal SD relates to the di f ference between the average sensor signal AS and the second sensor signal S , wherein the average sensor signal AS relates to an average of previous sensor signals detected by the fiber optic sensor 21 for the measurement segment 22 before detecting the first sensor signal F . The evaluation unit 25 comprises an alarm unit 28 that is configured to provide an alarm signal AL for the case that a confidence condition is ful filled, wherein the confidence condition requires at least that the first di f ference signal FD and the second di f ference signal SD are each larger than a predefined threshold signal . Optionally, the device 24 comprises a counter 23 that is configured to be incremented once or twice for each di f ference signal FD, SD being larger than the threshold signal and decremented once for each di f ference signal FD, SD being smaller than the threshold signal .

Figure 3 shows a fiber optic sensor 21 arranged along a railway track 20 . The fiber optic sensor 21 comprises an optical fiber 30 that extends along the railway track 20 . The optical fiber 30 comprises at least one measurement segment 22 . Furthermore , fiber optic sensor 21 comprises a detection component 29 . The detection component 29 is connected with the device 24 for detecting anomalies along a railway track 20 .

With figure 4 another exemplary embodiment of the method for detecting anomalies along a railway track 20 is described . A counter 23 is incremented twice for each di f ference signal FD, SD being larger than the threshold signal and decremented once for each di f ference signal FD, SD being smaller than the threshold signal . Thus , the counter 23 is incremented at least one more time for each di f ference signal FD, SD being larger than the threshold signal than for each di f ference signal FD, SD being smaller than the threshold signal . On the x-axis the time is plotted and on the y-axis the value of the counter 23 is plotted.

In figure 4, at first a first sensor signal F is detected. The first difference signal FD is larger than the threshold signal. Therefore, the counter 23 is incremented twice. In a next step, a third sensor signal T is detected. A third difference signal TD, which relates to the difference between the average sensor signal AS and the third sensor signal T is smaller than the threshold signal for each third sensor signal T. Thus, the counter 23 is decremented once. In a next step, a second sensor signal S is detected. The second difference signal SD is larger than the threshold signal. Therefore, the counter 23 is incremented twice. The confidence condition further requires that the counter 23 reaches three counts. This is achieved after detecting the second sensor signal S. Consequently, after detecting the second sensor signal S, the alarm signal AL is provided.

With figure 5 another exemplary embodiment of the method for detecting anomalies along a railway track 20 is described. The confidence condition further requires that between detecting the first sensor signal F and detecting the second sensor signal S less than three third sensor signals T are detected by the fiber optic sensor 21 for the measurement segment 22, wherein a third difference signal TD, which relates to the difference between the average sensor signal AS and the third sensor signal T, respectively, is smaller than the threshold signal for each third sensor signal T. On the x-axis the time is plotted. In figure 5 , at first three third sensor signals T are detected . Therefore , the confidence condition is not ful filled . At next , a first sensor signal F is detected . Afterwards , three third sensor signals T are detected . Therefore , again the confidence condition is not ful filled . At next , a first sensor signal F is detected . In a next step, a third sensor signal T is detected . Afterwards , a second sensor signal S is detected . Consequently, the confidence condition is ful filled and an alarm signal AL is provided .

With figure 6 the principle of fiber optic sensing is shown . As in figure 3 , the optical fiber 30 extends along the railway track 20 . The detection component 29 is connected with the optical fiber 30 . It is schematically shown that on the railway track 20 a rail vehicle 32 can move . The diagram in the top part of figure 6 shows backscattered signals detected by the fiber optic sensor 21 . On the x-axis the distance along the optical fiber 30 is plotted and on the y- axis the amplitude of the backscattered signals is plotted . At a first position xl along the optical fiber 30 a person 31 is walking . Thus , the amplitude of the backscattered signals is higher at the first position xl than at other positions where no mechanical vibrations occur . At a second position x2 along the optical fiber 30 a rail vehicle 32 is moving . Thus , the amplitude of the backscattered signals is higher at the second position x2 than at the first position xl where less mechanical vibrations occur .

With figure 7 it is shown how the principle of fiber optic sensing is employed in the method for detecting anomalies along a railway track 20 . Figure 7 shows a similar situation as in figure 6 . The optical fiber 30 extends along the railway track 20 . The detection component 29 is connected with the optical fiber 30 . The diagram in the top part of figure 7 shows backscattered signals detected by the fiber optic sensor 21 . On the x-axis the distance along the optical fiber 30 is plotted and on the y-axis the amplitude of the backscattered signals is plotted . At a first position xl along the optical fiber 30 a rail vehicle 32 is moving . Thus , the amplitude of the backscattered signals is higher at the first position xl than at other positions where no mechanical vibrations occur . At a second position x2 along the optical fiber 30 another rail vehicle 32 is moving . Below the position where the rail vehicle 32 is moving, there is a track defect 33 , this means the rail of the railway track 20 has a defect . A track defect 33 is one example for an anomaly along the railway track 20 . The backscattered signal detected during the passage of the rail vehicle 32 over the track defect 33 is di f ferent from the average sensor signal AS for the same position . In this example , the amplitude of the backscattered signal is increased at the position of the track defect 33 . This can lead to a di f ference signal FD, SD being larger than the threshold signal . In this way, the anomaly can be identi fied .

Figures 8 and 9 show exemplary embodiments of providing visual indications .

In Figure 8 , a heatmap is shown as a 2-dimensional plot . On the x-axis , di f ferent channels are given, each channel representing a measurement segment along a railway track . The y-axis represents time . A median value of all di f ference signals for a given timeframe is calculated and an STI ( Sensonic Track Index ) value is computed therefrom, where the STI value represents how many standard deviations the value is away from an average value without any event . Di f ferent STI values are represented by different respective hatching as shown on the right side of Figure 8. A very high STI value for multiple measurements and over a certain number of neighbouring measurement channels, as shown in the middle of Figure 8, visually indicates a broken rail in this example.

In Figure 9, a 1-dimensional plot of an example of providing visual indications of a broken rail is shown. In this plot, the x-axis represents date and time, whereas the y-axis shows the STI value. Each measurement segment, or channel, gets a different line type. All displayed pixels are difference signals, and the line is formed by the median value of multiple difference signals of the same segment over time. A rise of the lines, meaning higher median STI values over time, visually indicates a broken rail with increasing impact over time. The line type code of the different measurement segments or channels is shown on the right side of Figure 9.

Reference numerals

20 railway track

21 fiber optic sensor

22 measurement segment

23 counter

24 device for detecting anomalies along a railway track

25 evaluation unit

26 detection unit

27 subtraction unit

28 alarm unit

29 detection component

30 optical fiber

31 person

32 rail vehicle

33 track defect

AL alarm signal

AS average sensor signal

FD first di f ference signal

SD second di f ference signal

F first sensor signal

S second sensor signal

S 1-S5 steps

STI Sensonic Track Index xl first position x2 second position