KR20230085521A - railway and train monitoring system - Google Patents

Abstract

The present invention relates to a railway track and vehicle monitoring system, which includes a light source unit that emits pulsed light, an optical circulator that outputs input pulsed light to a sensing unit and outputs reversely scattered light to a detection unit, and sensing A sensing optical fiber, one end of which is connected to the line and installed on the line, an optical detector that receives and processes the light output from the detector, and a signal processor that calculates the position and speed information of the vehicle on the line from the signal output from the optical detector. The signal processing unit performs preprocessing on the collected signals output from the photodetector and restores the attenuation; A vehicle candidate area detection module that calculates a weight-based density based on the basis and detects an area where vibration signals are concentrated as a railway vehicle candidate area, and a Gaussian filter is applied to the vehicle candidate area to determine the length of the railway car from an area that is equal to or greater than the threshold value. A vehicle length calculation module for calculating and an object dynamic management module for determining the vehicle length information calculated in the vehicle length calculation module as an object and calculating attribute information including speed of the object.

Description

Railway track and vehicle monitoring system {railway and train monitoring system}

The present invention relates to a railroad track and vehicle monitoring system, and more particularly, to a railroad track and vehicle monitoring system capable of providing operation information including the state of a railroad track and the current location of a vehicle.

In order to detect the position on the track of a train, which is a vehicle running on the track of a railroad, conventionally, an electric circuit method is mainly used.

This electrical circuit method is a method of detecting trains by configuring an electrical circuit on the train tracks, and since electrical facilities must be continuously installed along the train tracks, not only does the installation cost increase, but also maintenance and inspection of electrical facilities The disadvantage is that additional costs are incurred for

In addition, the electric circuit method has a technical limitation in that the position of the train can be confirmed only within a range of about 1 km.

In order to improve these problems, a device for calculating the track condition and the position and speed of a train using backscattered light generated from an optical cable installed along the track is proposed in Korean Patent Registration No. 10-2124218.

However, the apparatus cannot provide train length information, and there is a need for a method of increasing measurement accuracy by compensating for attenuation of received scattered light as the length of the track increases.

The present invention was devised to solve the above requirements, and a railroad track and vehicle monitoring system capable of increasing measurement accuracy by calculating and providing length information of a vehicle running on a track and compensating for attenuation of received scattered light. Its purpose is to provide

In order to achieve the above object, a railway track and vehicle monitoring system according to the present invention includes a light source unit for emitting pulsed light; an optical circulator for outputting pulsed light emitted from the light source unit and inputted through an input terminal to a sensing terminal, and outputting scattered light traveling backward from the sensing terminal to a detection terminal; a sensing optical fiber, one end of which is connected to the sensing end, and installed on or around a track of a railway to be monitored; a photodetector for receiving and processing the light output from the detector; and a signal processing unit which controls the driving of the light source unit and calculates the position and speed information of the vehicle on the track from the signal output from the light detection unit, wherein the signal processing unit outputs the collected signal from the light detection unit. a pre-processing module for restoring attenuation by performing pre-processing; a binarization processing module for generating binarized data based on a threshold value (Vnoise), which is a noise level, for the signal generated through the preprocessing module; a vehicle candidate area detection module that calculates a weight-based density based on the binarization data generated by the binarization processing module and detects an area where vibration signals are concentrated as a railway vehicle candidate area; a vehicle length calculation module for calculating a railway vehicle length from an area equal to or greater than the threshold by applying a Gaussian filter to the vehicle candidate area determined by the vehicle candidate area detection module; and an object dynamic management module that determines the vehicle length information calculated by the vehicle length calculation module as an object and calculates attribute information including speed of the object.

According to the railway track and vehicle monitoring system according to the present invention, the length information of the vehicle operating on the track is calculated and provided, and the attenuation of the received scattered light is compensated for to increase the measurement accuracy for vehicle tracking.

1 is a diagram showing a track and vehicle monitoring system of a railway according to the present invention;
2 is a perspective view showing an installation example of the sensing optical fiber of FIG. 1;
3 is a view showing an installation example of a sensing optical fiber according to another embodiment of the present invention;
4 is a flow diagram showing a signal processing process of the signal processing unit of FIG. 1;
5 and 6 are graphs showing tracking states of railroad vehicles measured through the track and vehicle monitoring system of the railroad according to the present invention.

Hereinafter, a railway track and vehicle monitoring system according to a preferred embodiment of the present invention will be described in more detail with reference to the accompanying drawings.

1 is a diagram showing a track and vehicle monitoring system of a railroad according to the present invention.

Referring to FIG. 1, a railway track and vehicle monitoring system 100 according to the present invention includes a light source 110, a waveform generator 120, an optical circulator 130, a sensing optical fiber 150, and an optical detector 160. , a signal processing unit 180 is provided.

The light source 110 outputs pulsed light corresponding to the waveform signal output from the waveform generator 120 .

Here, the light source 110 and the waveform generator 120 correspond to the light source unit.

The optical circulator 130 outputs the pulsed light emitted from the light source 110 and inputted through the input terminal 130a to the sensing terminal 130b, and transmits scattered light that proceeds inversely from the sensing terminal 130b to the detection terminal 130c. ) is output as

The sensing optical fiber 150 has one end connected to the sensing end 130b of the optical circulator 130 and is installed on the track 10 of the railway to be monitored. As shown in FIG. 2 , the sensing optical fiber 150 may be installed along the extension direction in the neck portion of the line 10 formed in an 'I' shape.

Alternatively, of course, the sensing optical fiber 150 may be installed along the line 10 around the line.

As an example, as shown in FIG. 3, the sensing optical fiber 150 is installed along a passage in the conduit 50 built along the track 10 parallel to the subgrade 15 on which the track 10 is supported. etc. can be installed in various ways.

The light detector 160 receives and processes the light output from the detector 130c of the optical circulator 130 . The light detector 160 includes a filter (not shown) for filtering light in a band corresponding to Rayleigh scattered light with respect to the scattered light output from the detection terminal 130c of the optical circulator 130, and electrically converts the light output from the filter. It can be built as a photodetector that outputs as a signal.

The signal processing unit 180 controls the driving of the waveform generator 120 of the light source unit to generate pulsed light, and the object including the position and speed information of the vehicle on the line 10 from the signal output from the light detection unit 160 Attribute information of is calculated, and the calculated attribute information is output through the output unit 190. Here, the output unit 190 may be a display unit that displays attribute information or a communication interface that transmits data to a receiving device through a communication network.

Meanwhile, the signal processing unit 180 includes a preprocessing module 181, a binarization processing module 182, a vehicle candidate area detection module 183, a vehicle length calculation module 184, and an object dynamic management module 185.

The pre-processing module 181 restores the attenuation by performing pre-processing on the signals output from the photodetector 160 and collected. The pre-processing module 181 restores the distance attenuation to the same size level in all sections, since noise is strong and long distances must be monitored due to the nature of the railroad tracks 10 and vehicles running through the railroad tracks 10, that is, trains. Performs preprocessing to compare signals with .

The preprocessing and decay restoration process of the preprocessing module 181 will be described below.

First, the pre-processing module 181 generates a vibration signal through Equation 1 below from the Rayleigh backscattering signal generated in the sensing optical fiber 150 and collected over time from the photodetector 160.

는 레일레이 역산란 신호의 크기이고, zi는 미터단위를 적용하며 선로(10)의 i번째 거리지점이며 , tj는 j번째 시간이다.here,

is the magnitude of the Rayleigh backscattering signal, zi is the i-th distance point of the line 10 in meters, and tj is the j-th time.

는 진동신호로서

를 진동의 크기로 변환한 신호의 크기를 나타낸다. t_k는 진동신호 기준시간이며 이동평균과 인접차를 계산하는 시간이고, k는 기준시간 색인이다. 또한, M은 이동평균 계수이고, 바람직하게는 5~20사이의 값을 사용한다.also,

is the vibration signal

represents the magnitude of the signal converted to the magnitude of vibration. t _k is the reference time of the vibration signal, the time for calculating the moving average and the adjacent difference, and k is the reference time index. In addition, M is a moving average coefficient, and preferably uses a value between 5 and 20.

의 크기(파워)를 일정시간 누적하여 평균(Vzi)을 아래의 수학식 2에 의해 계산한다.The vibration signal is

The magnitude (power) of is accumulated over a certain period of time, and the average (Vzi) is calculated by Equation 2 below.

V(zi) has a characteristic of decreasing exponentially to zi by the attenuation coefficient of the sensing optical fiber 150.

지점에서의 평균파워크기

,

계산한다. 여기서 z_st는 센싱광섬유(150)의 시작 거리지점이고, z_en은 센싱광섬유(150)의 끝 거리지점이다. 또한

,

는 각각 z_st, z_en 지점의 거리-시간평균 진동크기로서, z_st, z_en 주변영역에서 시간평균 진동신호 V(zi)가 갖는 거리평균 크기이고 아래의 수학식3으로 표현된다. Next, the two points

Average power at point

,

Calculate. Here, z _st is a starting distance point of the sensing optical fiber 150, and z _en is an end distance point of the sensing optical fiber 150. also

,

is the distance-time average vibration magnitude of z _st and z _en points, respectively, and is the distance average magnitude of the time average vibration signal V(zi) in the area around z _st and z _en , and is expressed by Equation 3 below.

Here, A is the number of distance points for which an average is to be obtained, and is the width of an observation area for which a distance average is to be obtained.

Next, an estimated damping coefficient (α) is obtained, and a damping recovery curve is applied therefrom.

Damping Restoration Curve

Thereafter, the vibration signal subjected to damping restoration is calculated by Equation 5 below.

는 감쇠복원 거리별 시간평균 진동신호 이며, 거리별 시간평균 진동신호 V(zi)에 감쇠곡선

를 곱하여 생성하며, 감쇠복원된 거리별 시간평균 진동의 크기에 해당한다. here,

is the time-averaged vibration signal for each distance of attenuation restoration, and the decay curve for the time-average vibration signal V(zi) for each distance

It is generated by multiplying , and corresponds to the magnitude of the time-averaged vibration for each distance that has been restored.

The binarization processing module 182 generates binarized data based on a threshold value (Vnoise), which is a noise level, for the signal generated to be attenuated and restored through the preprocessing module 181.

를 잡음수준인 문턱값 V_noise 기준으로 이진화 처리한다.The binarization processing module 182 is a vibration signal damped and restored by Equation 6 below.

is binarized based on the threshold value V _noise , which is the noise level.

Here, Bo(zi) is a binarized signal for presence/absence of vibration in the time domain, and has a value of 0 or 1 by binarizing the presence/absence of vibration for each distance. That is, 1 means that a vibration signal exists at the point zi, and 0 means that there is no vibration signal.

In addition, V _noise is a time-average vibration amplitude threshold value for each distance applied to determine whether vibration occurs, and is preferably set as an average value of a region where no signal exists after the end point of the sensing optical fiber 150.

를 주파수 변환한 후 특징값을 추출하고 기계학습을 통해 이진화결과를 도출하도록 구축될 수 있다.Alternatively, the binarization processing module 182 is a vibration signal

It can be built to extract feature values after frequency transforming and derive binarization results through machine learning.

를 주파수 변환한 거리별 진동 주파수 신호인 For this, first

is a frequency-converted vibration frequency signal for each distance.

라고 할 때 주파수영역 특징값

)을 산출한다.

When the frequency domain feature value

) is calculated.

,

,

는 주파수영역 특징값이다.here,

Is a frequency domain feature value.

The feature value maximizing the statistical independence of the normal signal on the vibration signal can be extracted through Principle Component Analysis, Independent Component Analysis, T-distributed Stochastic Neighbor Embedding, etc.

의 구분선을 기준으로 진동신호와 정상신호를 분류하고, Support Vector Machine, 클러스터링 K-Mean Clustering으로 이진화 결과인 B1(zi)를 아래와 계산하도록 구축될 수 있다. 여기서,

는 주파수영역 진동 구분선으로 주파수영역에서 진동여부를 판단하기 위해 적용된 것이다. 또한, B1(zi)는 주파수영역 진동유무 이진화 신호다.In addition, as a machine learning model for distinguishing vibration signals,

It can be constructed to classify vibration signals and normal signals based on the dividing line of , and calculate B1 (zi), the result of binarization with Support Vector Machine and clustering K-Mean Clustering, as follows. here,

is a frequency domain vibration division line, which is applied to determine whether vibration occurs in the frequency domain. Also, B1(zi) is a binarization signal with/without vibration in the frequency domain.

Here, B _final (zi) is a final vibration presence/absence binarization signal, which is a vibration presence/absence binarization signal for determining final vibration by combining Bo(zi) and B1(zi).

The vehicle candidate area detection module 183 calculates a weight-based density based on the binarization data generated by the binarization processing module 182 to detect an area where vibration signals are concentrated as a railway vehicle candidate area.

The vehicle candidate area detection module 183 calculates the density of vibration events within a certain length by Equation 8 below in consideration of weights for each distance.

Here, D(zi) is the vibration density signal at point zi and represents the density of the binarized signal with and without vibration. Also, J is the weight application length. The weighted length is determined to include the length of 100 to 400m of the railroad car. As an example, if the length of a railway vehicle is based on 100 m, the following conditions are determined to be satisfied.

wj is 1 when j = 0 and is a Gaussian set to have a form in which the value decreases as j moves away from 0, and can be expressed by Equation 9 below.

Here, σ is the standard deviation. σ is set to J/3.

내에서 합산하여 기설정한 철도 차량 후보영역 문턱값(Dth)보다 크면 해당 영역은 철도차량이 존재하는 영역으로 산정한다.On the other hand, the calculated binarization signal density (D(zi)) is used as the candidate region range.

If it is greater than the threshold value (Dth) of the predetermined railway vehicle candidate area by summing up within the range, the corresponding area is calculated as an area where a railway vehicle exists.

The candidate region of the railway vehicle is determined when the sum of D(zi) vibration density signals within the W width is greater than Dth. As an example, W is set to 1000 m and Dth is set to 200.

z _max is the position of maximum vibration density. This is the point where the D(zi) vibration density signal is the largest within the candidate region of the railway vehicle.

The vehicle length calculation module 184 applies a Gaussian filter to the vehicle candidate region determined by the vehicle candidate region detection module 183, and calculates the railroad car length from a region equal to or greater than a threshold value.

후보영역내에서 최대값이 있는 위치를 검출한다. The vehicle length calculation module 184 is

The position with the maximum value in the candidate region is detected.

Here, z _max is the position of the maximum vibration density, and is the point where the D(zi) vibration density signal is the largest in the candidate region of the railway vehicle.

을 D(zi)값을 기준으로 계산하고, D(zi)가 설정된 일정 값인 C 이상이 되는 영역을 기차의 길이내라고 산정하고,

인 가장자리 값을 끝지점으로 계산한다.Next, it is divided into both sides based on zmax, and the left and right end points of the train

is calculated based on the value of D (zi), and the area where D (zi) is greater than or equal to C, which is a set constant value, is calculated as within the length of the train,

Calculate the edge value as the end point.

Here, C is the railroad car length detection threshold value, and is set to 5 to 40% of the D (z _max ) value.

Here, L is a value applied as a range for detecting the length of a railroad car, and is set to a value that can sufficiently include the length of a typical railroad car. As an example, L applies a value between 300 and 500 m.

은 잡음에 의해 일시적으로 검출된 정보일 수 있기 때문에, 일정시간 관찰하여 철도차량의 이동형태로 추정되는 길이 정보를 제외한 잡음신호를 제거하는 것이 바람직하다. Also, detected in the previous step

Since may be information temporarily detected by noise, it is desirable to observe for a certain period of time and remove the noise signal except for the length information estimated by the moving shape of the railroad car.

The object dynamic management unit 185 determines the vehicle length information calculated by the vehicle length calculation module 184 as an object, and calculates attribute information including the current location and speed of the object.

The object dynamic management unit 185 creates a new railroad car object when the length of the railroad car is detected over a certain period of time, and calculates the left end position, right end position, speed, and acceleration of the railroad car object using a tracking filter, to track As the tracking filter, a Kalman filter, a particle filter, an unscented Kalman filter, and the like may be applied.

In addition, the object dynamic management unit 185 destroys the object and ends the tracking when the railroad car object is not observed for a certain period of time or longer.

As shown in FIG. 4, the signal processing unit 180 collects backscattered signals from the photodetector 160 (step 210), and restores pre-processing and attenuation to the collected signals (step 220). Thereafter, binarized data on the presence or absence of vibration for each distance of the sensing optical fiber 150 is generated (step 230), and a railroad car candidate area is detected therefrom (step 240), and a railway car length is detected from the selected railway car candidate area. (step 250). Thereafter, after going through an object creation process of registering the railroad car whose length is determined as an object, tracking information for the created object is created and updated (step 260).

Meanwhile, results obtained by applying the present monitoring system 100 to an actual railway are shown in FIGS. 5 and 6 .

For demonstration experiments, this monitoring system 100 was installed in a communication room in Osong Station, and about 46 km of communication lines buried in pipelines along the tracks were connected by applying sensing optical fibers. The Osong-Gongju section route is operated by high-speed lines such as KTX and SRT. 5 and 6, the horizontal axis represents the distance from Osong Station, and the vertical axis represents the observation time.

In FIGS. 5 and 6, the parts marked with dots are the results when the position of the railroad car is detected based on the vibration level, and the line marked with a solid line is the railroad car tracking line calculated and obtained by the present monitoring system 100. am.

As can be seen in FIGS. 5 and 6, only points mapped based on the magnitude of vibration have a limit in that the location of the railway vehicle cannot be properly identified due to noise or sections where vibration spreads, such as tunnels and bridges. In addition, marked points may occur due to external vibration factors such as movement of a vehicle along a track, and there may be many cases in which they are erroneously detected as railroad vehicles.

On the other hand, the railroad vehicle tracking line calculated by the monitoring system 100 and mapped as a solid line in FIGS. 5 and 6 shows that it is possible to more precisely identify and track the location of the vehicle while excluding the influence of noise.

In this way, the present monitoring system 100 registers the railway vehicle as an object through the determination process as described above for the signal received through the photodetector 160, and then the tracking line mapped for the reception time and location is the distance It can be seen that the object can be precisely tracked both when the vehicle is moving in a direction in which the distance is moving away and when the vehicle is moving in a direction in which the distance is getting closer.

According to the railroad track and vehicle monitoring system described above, the length information of the vehicle running on the track is calculated and provided, and the attenuation of received scattered light is compensated for to increase measurement accuracy for vehicle tracking.

110: light source 120: waveform generator
130: optical circulator 150: sensing optical fiber
160: light detection unit 180: signal processing unit

Claims (6)

a light source unit that emits pulsed light;
an optical circulator for outputting pulsed light emitted from the light source unit and inputted through an input terminal to a sensing terminal, and outputting scattered light traveling backward from the sensing terminal to a detection terminal;
a sensing optical fiber, one end of which is connected to the sensing end and installed on or around a track of a railway to be monitored;
a photodetector for receiving and processing the light output from the detector;
A signal processing unit for controlling driving of the light source unit and calculating vehicle position and speed information on the track from a signal output from the light detection unit;
The signal processing unit
a pre-processing module for restoring attenuation by performing pre-processing on the signals output and collected from the photodetector;
a binarization processing module for generating binarized data based on a threshold value (Vnoise), which is a noise level, for the signal generated through the preprocessing module;
a vehicle candidate area detection module that calculates a weight-based density based on the binarization data generated by the binarization processing module and detects an area where vibration signals are concentrated as a railway vehicle candidate area;
a vehicle length calculation module for calculating a railway vehicle length from an area equal to or greater than the threshold by applying a Gaussian filter to the vehicle candidate area determined by the vehicle candidate area detection module;
An object dynamic management module that determines the vehicle length information calculated by the vehicle length calculation module as an object and calculates attribute information including speed for the object; railway track and vehicle monitoring system characterized in that it is provided. The method of claim 1, wherein the preprocessing module

Generates a signal attenuated by

Is the time-averaged vibration signal for each damping restoration distance,

where N is the accumulated time for obtaining the average power, V(zi) corresponds to the time average vibration signal for each distance,

ego,

is,

,

is the distance-time average vibration amplitude between the starting distance point (z _st ) of the sensing optical fiber and the end distance point (z _{en )} of the sensing optical fiber,

,

A railway track and vehicle monitoring system, characterized in that the condition of A is satisfied, and A is the number of distance points to obtain an average. The method of claim 2, wherein the binarization processing module
The damped restored vibration signal

The threshold value (Vnoise) compared to is applied to the average value of the area where the signal does not exist after the end point of the sensing optical fiber 150. Railway track and vehicle monitoring system. The method of claim 2, wherein the binarization processing module
A railway track and vehicle monitoring system, characterized in that the vibration signal is converted into a frequency domain to calculate a feature value, and the calculated feature value is classified through a machine learning model to generate a binary signal. The method of claim 1, wherein the vehicle candidate area detection module
If the binarization signal density of the presence or absence of vibration is calculated from the binarization data calculated by the binarization processing module, and the calculated binarization signal density is greater than the threshold value (Dth) of the railway vehicle candidate area set within the range of the candidate area, the corresponding area is a railway vehicle. A railway track and vehicle monitoring system characterized in that it is calculated as an existing vehicle candidate area. 6. The railway track according to claim 5 , wherein the vehicle length calculation module calculates the length of the rolling stock from a region that is equal to or greater than a threshold value by applying a Gaussian filter to the candidate vehicle region determined by the vehicle candidate region detection module. and vehicle monitoring systems.

Images