WO2017207830A1

WO2017207830A1 - Method and system for detecting and identifying rail vehicles on railways and warning system

Info

Publication number: WO2017207830A1
Application number: PCT/ES2016/070420
Authority: WO
Inventors: Enrique VALVERDE AGUILAR; Eduardo BERTRÁN ALBERTI
Original assignee: Agrupación Guinovart Obras Y Servicios Hispania, S.A.
Priority date: 2016-06-03
Filing date: 2016-06-03
Publication date: 2017-12-07
Also published as: ES2712661A1; ES2712661B1

Abstract

The invention relates to a method that uses at least one triaxial sensor disposed on a railway, which continuously captures a signal corresponding to a vibration generated by a rail vehicle travelling along any type of railway. Subsequently, the discrete Fourier transform is applied to the captured signal and the discrete spectrum is obtained. Following this, the discrete spectrum is correlated with a database that allows the type of rail vehicle travelling along the railways to be identified and the railway along which same travels to be reported, with the option of generating different actions (types of reports, generation of warnings and alarms) according to the type of railway vehicle detected. The method also uses a redundancy between two different technologies, an analogue subsystem and a digital subsystem, increasing certainty in the detection of the railway vehicle.

Description

METHOD AND SYSTEM FOR DETECTION AND IDENTIFICATION OF RAILWAY VEHICLES ON RAILWAYS AND WARNING SYSTEM

DESCRIPTION

OBJECT OF THE INVENTION

The object of the present invention relates to a method and a system for detecting and identifying railway vehicles on railway tracks and a warning system. The method, in addition to identifying the type of railway vehicle (ave train, alvia train, freight train, commuter train, construction machine, locomotive, etc.) that is circulating on the tracks, also informs whether the route by which This railway vehicle is circulating is the main road (in which the triaxial sensors that capture the vibratory signal at the passage of the rail vehicle are placed) or the adjacent tracks. This helps prevent situations that may pose a risk to people when maintenance work or works are being carried out on the roads through which the rail vehicle is traveling.

Find special application in the field of railway sector.

TECHNICAL PROBLEM TO BE RESOLVED AND BACKGROUND OF THE INVENTION

Patent document US 20150251674 A1 presents a method of detection by an acoustic generator placed in a mobile device (train), which can be adjusted with respect to its frequency spectrum for the identification of the mobile device by detecting the backscattered light between A small and finite group of frequencies. It is an active identification method, since it is the train itself that transmits a "signature" (discrete set of frequencies), which are detected as the vehicle passes, recognizing and identifying it. However, a spectral analysis is not properly done in a frequency range as in the present invention, but is only detected if the discrete set of frequencies received is one or the other to recognize the type of vehicle. Obviously this requires good coordination between the infrastructure manager (tracks) and the rail operator.

However, the method of detecting and identifying railway vehicles on railway tracks of the present invention employs at least one triaxial sensor (accelerometric or microphonic type) that continuously captures the generated vibration signal by a railway vehicle circulating on any type of railway track, to detect the approach of railway vehicles on the way to work areas, and identify both the type of railway vehicle that is circulating, and the route through which it is traveling, with option to generate different actions (types of reports, warnings or alarms) in accordance with the type of railway vehicle detected.

Additionally, the method takes advantage of a redundancy between two different technologies, an analog subsystem and another digital subsystem to increase security in the detection of the rail vehicle.

DESCRIPTION OF THE INVENTION

A first object of the invention relates to a method of detecting and identifying railway vehicles on railway tracks, a railway vehicle being understood as any type of rolling stock (vehicle equipped with wheels capable of driving on a railway track). This method, in a first scenario, where there is only one main road and there are no adjacent roads, comprises the following steps: generating an experimental database comprising a plurality of discrete spectra A corresponding to vibratory signals generated by a type of railway vehicle circulating on any type of railway track when the type of railway vehicle circulates on the main track on which at least one sensor (of triaxial or microphone accelerometric type) configured to pick up a vibratory signal from the passage of a railway vehicle is placed; continuously capture the signal corresponding to the vibration generated by the railway vehicle circulating on any type of railway track by means of the sensor disposed in said track, where the railway tracks can be formed by at least two rails, a set of fixings, and a type of support or a combination of a type of support and a type of sleeper; digitize the captured vibratory signal; calculating a discrete Fourier transform of said signal to obtain a discrete spectrum of the signal; make a first correlation of the discrete spectrum of the signal with each of the discrete spectra A of the database and obtain a correlation index for each correlation made; select from the database the discrete spectrum A that has a higher correlation index from among those correlation rates that are greater than a predefined threshold, when at least one of the correlation rates exceeds the predefined threshold, where said discrete spectrum A It indicates the type of railway vehicle and that the route on which it circulates is the main route; and the signal is monitored again. Additionally, the method is also able to analyze a second scenario, in which in addition to the main road, adjacent roads are also taken into account. In this case, it is necessary to add additional steps to the previous stages, so that it comprises expanding the experimental database with a plurality of discrete spectra B (B1, B2, ...) corresponding to a vibration generated by a type of railway vehicle traveling on any type of railway track when the type of railway vehicle is traveling on an adjacent track; make a second correlation of the discrete spectrum of the signal with each of the discrete spectra B (B1, B2, ...) of the database and obtain a correlation index for each correlation made; and select from the database the discrete spectrum B that has a higher correlation index from those correlation rates that are greater than a predefined threshold, when at least one of the correlation indexes exceeds the predefined threshold, where said discrete spectrum B indicates the type of railway vehicle and the route on which it is traveling.

Likewise, the method activates different alarm levels that indicate the type of railway vehicle detected and whether the route on which it is traveling is the main road or an adjacent road.

To obtain an even more complete detection method, it is possible to add two additional digital detection levels, which only warn of the presence or absence of a railway vehicle, and which are, a basic first level of digital detection (first digital detection) and very fast and a second level of digital detection (second digital detection) of a higher and more reliable level than this first level of detection. Thus, a first digital detection of the railway vehicle is performed, so that if the digitized signal has a duration greater than or equal to a predetermined time t1 and an amplitude greater than or equal to a predefined threshold u1 then a first digital level is activated alarm indicating that said first detection has been performed. Next, a second digital detection of the railway vehicle is also performed, by means of an analysis of a set of pre-established frequency subbands of the digitized signal, so that if all the samples of the set of frequency subbands exceed a threshold u2 predefined for a preset time t2 then a second digital alarm level is activated indicating that said second detection has been performed. For the generation of the database, the following steps have to be carried out: experimentally measure the vibratory signal generated by each type of railway vehicle circulating on any type of railway track when the railway vehicle circulates on the main road in which it has been placed the sensor and when the railway vehicle circulates on at least one adjacent track; digitize the vibratory signal; and calculate a discrete Fourier transform of the digitized vibratory signal, generating the discrete spectrum for each type of railway vehicle circulating on any type of railway track.

Parallel to the digital detection of the railway vehicle, an analog detection is also carried out, which seeks the speed and safety in the detection (only the presence or absence of the railway vehicle is reported), so that if the analog signal has a longer duration or equal to a preset time t3 and an amplitude greater than or equal to a predefined threshold u3 then an analog alarm level is activated indicating that said analog detection has been performed and the signal is monitored again; otherwise the signal is also monitored again.

To ensure proper operation, the method additionally transmits pilot tones to a receiving unit to report the status of the sensor, connections and links.

In the event that no discrete spectrum has been selected from the database due to not finding matches, the method stores the data corresponding to said unknown railway vehicle for further learning and recognition of new railway vehicles.

A second object of the invention relates to a system for detecting and identifying railway vehicles on railway tracks such that it comprises: at least one sensor arranged in the track configured to continuously capture a signal corresponding to a vibration generated by a railway vehicle circulating on any type of railway track; computational means configured to: store an experimental database comprising a plurality of signals and their corresponding discrete spectra, where each signal corresponds to a vibration generated by a railway vehicle circulating on any type of railway track; digitize the captured vibratory signal; calculate a discrete Fourier transform of said signal to obtain a discrete signal spectrum; correlate the discrete spectrum of the signal with each of the discrete spectra of the database that correspond to the same type of route through which the railway vehicle circulates, and obtain a correlation index for each correlation made; and select the discrete spectrum of the database that has a higher correlation index from those correlation indexes that are greater than a predefined threshold, when at least one of the correlation indexes exceeds the predefined threshold.

Additionally, the railway vehicle detection and identification system on railway tracks comprises an alarm level activation system when the railway vehicle is detected and / or identified.

A third object of the invention relates to a warning system comprising the detection and identification system of railway vehicles on railway tracks.

Therefore, the method of the present invention has the following advantages over current detection methods:

- Identify the type of railway vehicle in any scenario (different types of tracks) generating warnings or alarms accordingly.

- Recognizes by means of correlation techniques and spectral analysis, the type of railway vehicle and identifies whether it is approached by the main road where the sensors that capture the vibratory signal generated by the passage of the railway vehicle or by adjacent tracks are installed.

- The rail vehicle does not actively participate in its own identification as ID, RFID codes, or the transmission of other identifiers from the train are not used. Nor does it require installing generators in rail vehicles, this is a passive detection method. This achieves greater operational autonomy between infrastructure management, vehicle operation and supervision or maintenance tasks.

- Early detection of the rail vehicle without requiring the rail vehicle in front of the sensor.

- Redundancy between analog and digital subsystems to combine security with computing capacity and for better activation of fail-safe states. This redundancy, which covers from the detection of the passage of a railway vehicle to the generation of warnings, increases the reliability of the warning system. - Good operational independence between the infrastructure management entity, the railway operator and the maintenance companies.

BRIEF DESCRIPTION OF THE FIGURES

In order to complete the description and in order to help a better understanding of the features of the invention, this descriptive report is attached, as an integral part thereof, a set of drawings where, for illustrative and non-limiting purposes, it has been represented the next:

Figure 1.- Shows a flow chart of the method of the present invention for the case in which only the main road exists (there are no adjacent roads). It is observed how the digital subsystem comprises the first correlation of the third level of digital detection.

Figure 2.- Shows a flow chart of the method of the present invention for the case in which in addition to the existence of the main route, there are also adjacent routes. It is observed how the digital subsystem then comprises the first and the second correlation of the third level of digital detection.

Figures 3a and 3b.- They show a flow chart of the method of the present invention for the most complete case, in which in addition to the main road, there are also adjacent roads. It is observed how the digital subsystem comprises the three levels of digital detection, first level, second level and third level of digital detection (first and second correlation). Figure 3b is the continuation of Figure 3a.

Figure 4.- Shows an example of an experimental database, which contains parameters such as the type of railway vehicle, type of support and type of sleeper and pre-memorized patterns (discrete spectra) corresponding to different types of railway vehicles circulating through different ways.

Figure 5.- Shows the system devices that allow this method and the warning system to be carried out.

The following is a list of the references used in the figures:

1. Main way.

2. Adjacent way.

3. Sensor.

4. Computational media.

5. Receiving unit. 6. Warning system. PREFERRED EMBODIMENT OF THE INVENTION

A first object of the present invention describes a method of detecting and identifying railway vehicles on railway tracks comprising the following steps:

1. Continuously capture and monitor a signal (continuous measurements on the x, y, z axes) corresponding to a vibration generated by a railway vehicle traveling on any type of railway via sensors arranged at one or more points of interest at along the way. The sensors can be accelerometric or microphone type, preferably triaxial accelerometric sensors.

2. Amplify (with operational amplifiers or low frequency transistors) and filter the vibratory signal captured by the sensor to eliminate noise caused by other sources outside the approach of a railway vehicle on the track.

This first filtering is carried out with low-pass filters of cut-off frequency of the order of a few kHz (not more than 5 kHz, and typically of the order of kHz or 500 Hz, according to the accelerometric sensor used), preserving in the frequency band of interest, the information in amplitude and frequency.

3. After this first filtering, the vibrating signal captured is sent, in parallel, to two subsystems, an analog subsystem, which seeks the speed and safety in the detection of the rail vehicle (only the presence or absence of the rail vehicle is reported) and another digital subsystem, which, in addition to confirming said detection at different levels, identifies the type of railway vehicle and informs of the route through which said railway vehicle is traveling.

3.1 The analog subsystem is responsible for checking whether the vibrating signal has a duration greater than or equal to a preset time t3 and an amplitude greater than or equal to a predefined threshold u3, thus preventing short pulses and other noise (interference, non-ideal mechanical contacts, discontinuities, road bumps, work / maintenance machinery ...) can generate false alarms for the detection of a railway vehicle. If the analog vibrating signal has a duration greater than or equal to t3 and an amplitude greater than or equal to the threshold u3, then a first analog alarm level is activated, indicative of the presence of a railway vehicle. Subsequently, whether or not the rail vehicle has been detected, the signal is monitored again.

The processing is totally analog, so it does not depend on digital aspects, such as program counters, which in the face of strong electromagnetic interference (for example, construction machinery) could deceive the iterative execution of the algorithms (erroneous program counter jumps, partial erasure or destruction of memories, etc).

This analog subsystem comprises the following elements:

A low pass filter (can be RC type, and does not necessarily have to be active), with low thermal variation and environmental robustness, with low cutoff frequency, of the order of a few Hz (at most hundreds of Hz, although usually tens of Hz ). This filter acts as an integrator, since it needs a minimum of continuity in the vibrations. It is important to adjust the time constant of this filter: a slow adjustment may not fire on short vehicles or, simply, locomotives without wagons, while a quick adjustment may trigger for the mentioned false alarms. As indicative, they are considered time constants of the order of 2 to 15 seconds.

An envelope detector (preferably formed with a circuit consisting of diodes, resistors and capacitors) whose output will be to detect the envelope (amplitude) of the vibratory signals.

A level detector (comparator circuit type) connected to the output of the envelope detector, which activates the detection when its output exceeds a predefined trigger threshold. The detector is implemented as an analog comparator, whether based on transistors or operational amplifiers.

In the most basic case, the level detector only warns of the presence or absence of a railway vehicle. While in a more complete case, the level detector (with a window comparator) reports said detection with three levels of reliability (confidence): safe presence of railway vehicle, safe absence and uncertainty. To adjust the sensitivity of this level detector depending on the type of track, a precalibration is made between selectable values with a resistive divider and a switch (which can be mechanical or a resistance network adjustable by keyboard using CMOS technology switches).

A PCM modulator at the output of the level detector for transmission, modulated or in baseband (signal transmitted in the same frequency band in which it has been captured, without modulation), wired or transmitted by radio frequencies, to the information receiver on train detection (control center, maintenance brigades, etc.)

3.2 The digital subsystem (based on microcomputer, digital signal processor, FPGA or similar device), acts independently of the analog subsystem, complementing it, since in addition to corroborating the presence of the railway vehicle, it also identifies the type of railway vehicle that is circulating and informs whether the route through which said railway vehicle is traveling is the main route (in which the sensor (s) are placed) or adjacent tracks.

The digital subsystem comprises an analog / digital (A / D) converter of not less than 8 bits, for the digitalization of the captured vibratory signal and which follows a previous level conditioner (amplification to condition the output level of the sensors to the dynamic range of the A / D converters, to take advantage of their benefits, and pre-filtered to prevent the inevitable environmental noise or causes beyond the passage of a railway vehicle could mask future decisions), all with or without a sampling and maintenance device (S&H sample-and-hold subsystem, responsible for maintaining constant the signal of the sensors while the process of translating the analog domain into a digital word intelligible by the subsequent microprocessors), depending on the A / D converter technology used . The sampling rate must be a theoretical minimum of 10 kHz, preferably a minimum of 50 kHz, although this value is susceptible to a wide range of variation between Hz and kHz, depending on the accelerometric sensor used, of the resolution ( in "g", a unit that takes as a reference the acceleration of the desired gravity) and sensitivity, as some sensors offer different benefits by varying the sampling frequency (speed at which the processor acquires the samples (digital words) of the accelerometers).

This digital subsystem comprises three levels of detection of railway vehicles: a first level of basic and very fast digital detection, a second level of digital detection of a higher and more reliable level than the first level of detection and a third level of digital detection which is the one that in addition to detecting the railway vehicle, identifies the type of a railway vehicle and provides information on whether the passage of said railway vehicle is produced by the main road (in which the triaxial sensors are placed) or by the adjacent tracks. Knowing whether the rail vehicle is passing through the main road or adjacent roads is decisive, since in this way, in the case that the rail vehicle is passing through the adjacent tracks and not the main track, It is not necessary that people who are carrying out maintenance work on this road withdraw work equipment (machinery), beyond respecting the gauge, and can continue with their work.

A first level of digital detection, which comprises performing a first digital detection of the railway vehicle, so that if the digitized signal has a duration greater than or equal to a predetermined time t1 and an amplitude greater than or equal to a predefined threshold u1 then it is activated a first digital alarm level indicating that said first detection (presence of railway vehicle) has been performed. Subsequently, whether the first digital alarm level has been activated or not, a second level of digital detection would be evaluated.

In other words, this first level of detection is activated by recording activity maintained in the sensor for a period of time. If there is an output of amplitude greater than a predefined threshold of "g" in the sensor during a whole temporary window of long enough length (about 10 seconds) so that the passage of a railway vehicle is not confused with a vibration for other reasons , like tapping on the track.

This level also informs the direction of the approach of the railway vehicle (due to relative amplitudes between different temporary bursts of sampling, since the amplitude of vibration increases when the railway vehicle approaches the sensor, and decreases when moving away; of the degree of proximity of the railway vehicle and of the passing speed of the railway vehicle.

A second level of digital detection, which comprises performing a second digital detection of the railway vehicle, by means of an analysis of the presence of energy (of vibrations) at the exit of a bank of digital filters each adjusted to a different frequency band within a set of preset frequency subbands of the digitized signal, so that if all frequency subbands exceed a predefined threshold u2 for a preset time t2 then a second digital alarm level is activated indicating that said second detection has been made (presence of railway vehicle). Subsequently, whether the second digital alarm level has been activated or not, a third level of digital detection would be evaluated.

That is, this second level of detection detects vibrations around a minimum and predefined set of between 2 and 5 frequency subbands, allowing a certain degree of tolerances, to identify the vibrations corresponding to the approximation of railway vehicles of other types of vibrations. For this, it is monitored whether the outputs of digital filters type FIR (finite impulse response, more stable) or MR (infinite impulse response, shorter calculation time), pass-band type and adjusted to the frequencies that contain higher Information (of the order of 1 to 10 Hz) exceeds a predefined threshold u2 for a preset time t2 (approximately of the order of 2 to 15 seconds). If all the outputs of the digital filters exceed said predefined threshold u2 during the preset time t2, the corresponding alarm is activated.

For example, in the case of having a maximum of 5 frequency bands, if vibratory energy is detected that exceeds the threshold u2 during the time t2 in the 5 bands (bands activated), then the approaching railway vehicle alarm is activated. However, if vibratory energy is detected in the 5 bands but does not exceed the threshold u2 during time t2 or vibrational energy is detected that exceeds the threshold u2 during time t2 in one, two, three or four of the bands (se understand that if there is no vibratory energy in any of the 5 frequency bands, no railway vehicle is approached), then the filtering is repeated for a new frame of acceleration measurements.

A third level of digital detection, slower than the previous ones but with more information, which is the main object of the invention, which comprises correlating the spectral samples of the discrete spectrum of the signal that is obtained in real time as the vehicle passes railway with each of the discrete spectra of pre-memorized patterns in a database to identify the type of vehicle railway, and also identify the route along which it is circulating, and may be the main route (in which the triaxial sensors are placed) or adjacent ones. Additionally, a corresponding alarm level is activated indicating the type of railway vehicle detected and the route through which it circulates.

The database has been generated experimentally and comprises a plurality of discrete spectral patterns, where each discrete spectrum corresponds to a vibration generated by a particular type of railway vehicle circulating on any type of railway track, where the railway tracks are formed by a type of support (as is the case of the plate track) or a combination of a type of support and a type of crossbeam (as is the case of ballast with wooden or concrete sleepers).

For the generation of the database, the following steps have been carried out: on the one hand, experimentally measure the vibratory signal generated by each type of railway vehicle traveling on any type of track when the railway vehicle travels on the same track (main track ) on which the sensors have been placed, and on the other hand, experimentally measure the vibratory signal generated by each type of railway vehicle circulating on any type of track when the railway vehicle circulates along adjacent tracks (usually there will only be one adjacent track, however, there may be more than one adjacent track); digitize the vibratory signals; and calculate the discrete Fourier transform of said digitized vibratory signals, generating, on the one hand, the discrete spectra for each type of railway vehicle circulating on any type of track when the railway vehicle circulates on the same track in which they have been placed the sensors (discrete spectra A) and on the other hand, the discrete spectra for each type of railway vehicle circulating on any type of track when the railway vehicle is traveling on an adjacent track (discrete spectra B1 (adjacent track), B2 (adjacent track2) ), B3 (adjacent track3), etc).

The discrete spectra B (B1, B2, ...) differ from the discrete spectra A mainly in two aspects: its lower amplitude (the sensor measures the vibratory signal that comes from the adjacent paths and not from the main path in which the sensor is placed) and a greater attenuation of the high frequencies with respect to the low frequencies, given the low-pass behavior of the propagation of mechanical vibration waves on the ground. This third level of digital detection calculates a discrete Fourier transform of the vibratory signal captured by the sensor (s) to obtain the discrete spectrum of said signal. Subsequently, a first correlation of the discrete spectrum of the signal obtained with each of the discrete spectra patterns A of the database that correspond to different discrete spectral patterns of different railway vehicles circulating on the same type of track is made, said patterns having been evaluated on the main path, in which the sensors have been placed, so that a correlation index is obtained for each first correlation made. Next, the discrete spectrum A that has a higher correlation index (it is understood that there can be no correlation indexes with the same value) is selected from the database from those correlation indexes that are greater than a predefined threshold ( for example, correlation index threshold> 0.7), in this case, if this correlation index is found, an alarm level is activated which, in addition to identifying the type of railway vehicle being circulated, also identifies that the route through the one that circulates is the main route (where the sensors are installed) and the signal would be captured and monitored again. Obviously, the correlation index threshold can be modified according to different criteria.

However, in the event that none of the correlation indexes exceed the predefined threshold, a second correlation of the discrete spectrum of the signal obtained with each of the discrete spectra patterns B (B1, B2, ...) is made. ) of the database that correspond to different patterns of discrete spectra of different railway vehicles traveling on the same type of track, said patterns having been evaluated on adjacent tracks, so that a correlation index is obtained for each second correlation made. Next, the discrete spectrum B (B1, B2, ...) that has a higher correlation index is selected from the database among those correlation rates that are greater than a predefined threshold (for example, index threshold of correlation> 0.7), in this case, if said correlation index is found, an alarm level is activated which, in addition to identifying the type of railway vehicle that is traveling, also identifies the adjacent route through which it circulates, of such so it is not necessary for people who are carrying out maintenance work on the main road (in which the sensors are placed) to remove the work equipment placed on said main road, and the signal would be captured and monitored again. In the event that none of the correlation indices (of the first correlation and the second correlation) exceeds the predefined threshold, the signal would be re-monitored and a presence information of the unidentified railway vehicle type would be stored (if there is a railway vehicle ), for the management of a database that can be used, for example, for the subsequent learning and recognition of new railway vehicles. The stored information may contain data such as the hour and minute and / or the discrete spectrum of vibrations produced, or only its main parameters to reduce storage requirements.

The correlation operation involves comparing the sequences of samples corresponding to a vibratory burst with other sequences corresponding to type bursts that identify different situations, depending on whether the identification of the type of railway vehicle or the identification of the type of passageway is sought. The term correlate is emphasized because mathematically it allows comparison to be made regardless of the start time of each sequence, only by its form. However, in the present invention, although the temporary samples (those taken directly from the sensors) could be correlated, the term correlate refers to the correlation of spectral samples: this means making a fast Fourier transform (FFT type), which generates a sequence of discrete samples, each corresponding to a certain frequency (hence the name of spectral samples), and this is the sequence that is correlated with discrete spectrum patterns.

A particular example to better understand the invention is described below. Figure 4 shows a database (table) containing the discrete spectra of pre-memorized patterns in which there are 3 tracks (the main track and two adjacent tracks) where X, Y, Z, ... they represent different types of railway vehicles (for example Alvia, AVE, locomotive, goods, commuter trains, ...). Each railway vehicle generates a spectrum of vibrations depending on the type of track on which it is traveling, so there are several spectrum patterns (frequencies that are activated and amplitudes at the most significant frequencies) for each type of railway vehicle. The column relating to the discrete spectra A refers to the pre-memorized patterns corresponding to the discrete spectrum of different types of railway vehicles traveling on one type of track, when the passageway is the main track, on which the / the sensors Columns related to discrete spectra B (column B1 and B2) refer to the pre-memorized patterns corresponding to the discrete spectrum of the different types of railway vehicles circulating along the adjacent tracks, where the discrete spectra B1 correspond to the adjacent track and the discrete spectra B2 correspond to the adjacent track2.

The data that is obtained as each stage of the method is executed (presence of railway vehicles on track, types of railway vehicles, etc.) are stored locally (on a measurement basis) and transmitted by cable or radio to a receiving unit ( alarm center, centralized control rooms, maintenance brigades that are working on the affected roads, etc) that manages them. The different connections and links between the system elements can be made by cable or by wireless technology.

Additionally, the method comprises transmitting, continuously or discontinuously, pilot tones (fixed frequencies) or digital codes, to a receiving unit to inform (in fault tolerant applications) of the state of attention of the sensors and the reliability of the connections and of the links (radio or wired).

Figure 1 shows the method of the present invention represented by a flow chart for the case in which only the main road exists (there are no adjacent roads). On the one hand, the analog subsystem is observed, and on the other hand, and running in parallel, the digital subsystem comprising the first correlation of the third level of digital detection, where said first correlation allows to identify the type of railway vehicle and provides information on if the passage of said railway vehicle occurs through the main road (in which the triaxial sensors are placed) or not. Therefore, in this figure 1 neither the first level of digital detection nor the second level of digital detection is shown.

As shown in Figure 1, in step 100, the method continuously captures and monitors the vibratory signal. In step 101, the method performs noise filtering and level conditioning. Step 102 (analog subsystem) performs an analog detection of the railway vehicle and in parallel, step 1 10 (digital subsystem) performs a digital detection of the railway vehicle.

After step 102, the method performs envelope detection, filtering and comparison (step 103). In step 104 the method determines if a predefined threshold has been exceeded for a while. If this threshold is exceeded an alarm is activated indicative of the presence of a vehicle, step 105, and the method continues in step 100. If the predefined threshold is not exceeded for a period of time no railway vehicle presence detection alarm is activated and the method continues in step 100. After step 1 10, the method digitizes the captured vibratory signal (step 1 11) and then, in step 112, the discrete Fourier transform of said signal is calculated to obtain a discrete spectrum of the signal. In step 113 a first correlation of the discrete spectrum of the signal is made with each of the discrete spectra A of the database obtaining a correlation index for each correlation made, and it is verified (step 1 14) if the discrete spectrum of the signal coincides with some discrete spectrum A. If it coincides, the method activates an alarm (step 1 15) informing of the type of railway vehicle and that the route through which it circulates is the main route and returns to the stage 100; otherwise (if none match) it also returns to stage 100.

Figure 2 shows the method of the present invention represented by a flow chart for the case in which in addition to the main path, there are also adjacent paths. In this case, the digital subsystem comprises the first correlation and a second correlation of the third level of digital detection. Therefore, this figure 2 comprises the flowchart of figure 1, where in the event that no discrete spectrum A coincides with the discrete spectrum of the signal, instead of returning to step 100, a second one would be carried out correlation of the discrete spectrum of the signal with each of the discrete spectra B (B1, B2, ...) of the database (step 1 16) obtaining a correlation index for each correlation made, and is verified (step 1 17) if the discrete spectrum of the signal coincides with some discrete spectrum B (B1, B2, ...). If it coincides, the method activates an alarm (step 1 18) informing of the type of railway vehicle and the route through which it circulates and returns to stage 100; otherwise (if none match), the data corresponding to the unknown railway vehicle (if any) is saved (stage 1 19) and returns to stage 100.

Figure 3 shows the method of the present invention represented by another flow chart, where you can see, on the one hand, the analog subsystem (which is the same as in Figure 1 and 2) and on the other hand, and operating in parallel, the digital subsystem which in this case comprises the first level of digital detection, the second level of digital detection and the third level of digital detection, where the third level of detection comprises the first and the second correlation. The stages of the digital subsystem 210 of this figure 3 are listed below. In step 211 the method digitizes the captured vibratory signal and then, in step 212 is performed the first level of digital detection, where it is evaluated whether or not there is a railway vehicle detection (step 213). If there was a rail vehicle detection, the method activates a first digital alarm level (step 214) and goes to step 215, while if there has been no rail vehicle detection, it would go directly to step 215. In step 215 the second level of digital detection is performed, where it is also evaluated whether there is a detection or not of a railway vehicle (step 216). If there has been a railway vehicle detection, the method activates a second digital alarm level (step 217) and goes to step 218, while if there has been no railway vehicle detection it would go directly to step 218. From here , the method would execute the third level of digital detection (first and second correlation) and therefore, the same steps as the method of Figure 2, that is, in step 218, the discrete Fourier transform of said signal is calculated for Obtain a discrete signal spectrum. In step 219 a first correlation of the discrete spectrum of the signal is made with each of the discrete spectra A of the database obtaining a correlation index for each correlation made, and it is verified (step 220) if the discrete spectrum of the signal coincides with some discrete spectrum A. If it coincides, the method activates an alarm (step 221) informing of the type of railway vehicle and that the route through which it circulates is the main route and returns to stage 100 ; otherwise (if none match) a second correlation of the discrete spectrum of the signal is made with each of the discrete spectra B (B1, B2, ...) of the database (step 222) obtaining an index of correlation for each correlation made, and it is verified (step 223) if the discrete spectrum of the signal coincides with some discrete spectrum B (B1, B2, ...). If it coincides, the method activates an alarm (step 224) informing of the type of railway vehicle and the route through which it circulates and returns to stage 100; otherwise (if none match), the data corresponding to the unknown railway vehicle (if any) is saved (step 225) and returns to step 100.

An example of an experimental database is shown in Figure 4.

A second object of the invention describes a system comprising the sensors and computational means (computer with processor) configured to carry out the above method.

As shown in Figure 5, said system therefore comprises at least one sensor (3) arranged in the path configured to continuously capture and monitor a signal corresponding to a vibration generated by a railway vehicle traveling on any type of track; computational means (4) configured to: store an experimental database comprising a plurality of signals and their corresponding discrete spectra, where each signal corresponds to a vibration generated by a railway vehicle circulating in any way, digitize the captured vibratory signal , calculate a discrete Fourier transform of said signal to obtain a discrete spectrum of the signal, correlate the discrete spectrum of the signal with each of the discrete spectra of the database that correspond to the same type of path through which it circulates the railway vehicle, and obtain a correlation index for each correlation made, and select the discrete spectrum of the database that has a higher correlation index from those correlation indexes that are greater than a predefined threshold, when at least one of the correlation indexes exceeds the predefined threshold. The system additionally comprises an alarm level activation system when the railway vehicle is detected and identified.

A third object of the invention describes a warning system (6), represented in Figure 5, which comprises the system for detecting and identifying railway vehicles on railway tracks.

The present invention should not be limited to the embodiment described herein. Other configurations can be made by those skilled in the art in view of the present description. Accordingly, the scope of the invention is defined by the following claims.

Claims

1. Method for detecting and identifying railway vehicles on railway tracks characterized by the following stages:

to. generate an experimental database comprising a plurality of discrete spectra A corresponding to vibration signals generated by a type of railway vehicle circulating on any type of railway track when the type of railway vehicle circulates on a main track on which it is placed at least a sensor configured to capture a vibrating signal from the passage of a railway vehicle;

b. continuously capture the signal corresponding to the vibration generated by the railway vehicle traveling on any type of railway track using the sensor arranged on said track, where the railway tracks are formed by a type of support or a combination of a type of support and a type of mischief; c. digitize the captured vibration signal;

d. calculating a discrete Fourier transform of said signal to obtain a discrete spectrum of the signal;

and. perform a first correlation of the discrete spectrum of the signal with each of the discrete spectra A of the database and obtain a correlation index for each correlation carried out;

F. select from the database the discrete spectrum A that presents a higher correlation index among those correlation indices that are greater than a predefined threshold, when at least one of the correlation indices exceeds the predefined threshold, where said discrete spectrum A indicates the type of railway vehicle and that the track on which it circulates is the main track;

g. and return to b.

2. Method for detecting and identifying railway vehicles on railway tracks according to claim 1, characterized in that between stages f and g it comprises:

- expand the experimental database with a plurality of discrete spectra B corresponding to a vibration generated by a type of railway vehicle circulating on any type of railway track when the type of railway vehicle circulates on an adjacent track;

- perform a second correlation of the discrete spectrum of the signal with each of the discrete spectra B of the database and obtain a correlation index for each correlation carried out; and

- select from the database the discrete spectrum B that presents a higher correlation index among those correlation indices that are greater than a predefined threshold, when at least one of the correlation indices exceeds the predefined threshold, where said discrete spectrum B indicates the type of railway vehicle and the track on which it circulates.

3. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that an alarm level is activated that indicates the type of railway vehicle detected and whether the track on which it is traveling is the track. main or an adjacent road.

4. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that between stages c and d it comprises performing a first digital detection of the railway vehicle, so that if the digitized signal has a duration greater than or equal to at a pre-established time t1 and an amplitude greater than or equal to a predefined threshold u1, then a first digital alarm level is activated indicating that said first detection has been made.

5. Method for detecting and identifying railway vehicles on railway tracks according to claim 4, characterized in that it comprises performing a second digital detection of the railway vehicle, through an analysis of a set of pre-established frequency sub-bands of the digitized signal, so that if all the samples of the set of frequency sub-bands exceed a predefined threshold u2 during a preestablished time t2, then a second digital alarm level is activated that indicates that said second detection has been made.

6. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that the generation of the database comprises:

• experimentally measure the vibration signal generated by each type of railway vehicle circulating on any type of railway track when the railway vehicle circulates on the main track on which the sensor has been placed and when the railway vehicle circulates on at least one adjacent track;

• digitize the vibration signal; and, • calculate a discrete Fourier transform of the digitized vibration signal, generating the discrete spectrum for each type of railway vehicle circulating on any type of railway track.

7. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that parallel to step c an analog detection of the railway vehicle is carried out, so that if the analog signal has a duration greater than or equal to at a pre-established time t3 and an amplitude greater than or equal to a predefined threshold u3, then an analog alarm level is activated that indicates that said analog detection has been made and returns to b; Otherwise, b is executed.

8. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that the method comprises transmitting pilot tones to a receiving unit to report the state of the sensor, connections and links.

9. Method for detecting and identifying railway vehicles on railway tracks according to claims 1 or 2, characterized in that it comprises storing data corresponding to an unknown railway vehicle when no discrete spectrum has been selected from the database.

10. System for detecting and identifying railway vehicles on railway tracks to carry out the method defined in claims 1 to 9, characterized in that it comprises:

- at least one sensor (3) arranged on the track configured to continuously capture a signal corresponding to a vibration generated by a railway vehicle traveling on any type of railway track;

- computational means (4) configured to:

• store an experimental database comprising a plurality of signals and their corresponding discrete spectra, where each signal corresponds to a vibration generated by a railway vehicle traveling on any type of railway track;

• digitize the captured vibration signal;

• calculate a discrete Fourier transform of said signal to obtain a discrete signal spectrum;

• correlate the discrete spectrum of the signal with each of the discrete spectra in the database that correspond to the same type of track on which the railway vehicle circulates, and obtain a correlation index for each correlation made; and

• select the discrete spectrum from the database that presents a higher correlation index among those correlation indices that are greater than a predefined threshold, when at least one of the correlation indices exceeds the predefined threshold.

1 1. System for detecting and identifying railway vehicles on railway tracks according to claim 10, characterized in that it comprises a system for activating alarm levels when the railway vehicle is detected and identified.

12. System for detecting and identifying railway vehicles on railway tracks according to claim 10, characterized in that the sensor (3) is of the triaxial accelerometric type.

13. Warning system (6) characterized in that it comprises the system for detecting and identifying railway vehicles on railway tracks according to any one of claims 10 to 12.