EP2305532A1 - Method for automated synchronisation of rail position measurements - Google Patents

Abstract

The method involves determining measured values of measuring channels for calculated local sample values by an interpolation procedure. A reference measurement and measurements to be synchronized are divided into sections. An offset is determined from local offsets of all the sections, and is utilized for all the measuring channels. Measured values of the measurements to be synchronized are calculated at positions of the reference measurement, where measured values of the reference measurement and the measurements to be synchronized have same position co-ordinates.

Description

The invention relates to a method for automatically synchronizing track position measurements in railway traffic.

The railway infrastructure is the basis for a safe railway operation. To ensure the high degree of safety, extensive maintenance measures are necessary. The inspection of the track geometry is of particular importance here. Often the inspection of the track position is carried out with special measuring vehicles. The results obtained are compared with limit values. From the comparison repair measures are derived and carried out. The measurement results of the track position inspections are stored in databases. With special programs it is possible to display the measurement results of individual inspection trips. In this case, several results of different inspection trips of the same route section can be displayed at different inspection times. From this presentation, a person skilled in the art can follow the development of track deviations and plan and carry out medium and long-term maintenance measures. Since the results of different inspection runs of the same section of the route at different inspection times have been shifted locally up to 100 m, an automated evaluation and evaluation of the results is not possible. In addition, the local increments are not always constant, so that additionally creates a distortion of the measurement. This local offset and the distortion arise because the distance measurement is usually done by means of wheel pulses. Due to wear-related decrease in the wheel diameter, slip during braking and acceleration as well as due to friction and weather conditions, these deviations can not be avoided.

Disadvantages of the prior art:

Although it is possible in the current state of the art for the results of different inspection journeys to be compared manually, an automated evaluation and evaluation is not possible.

For an automatic evaluation and evaluation, the results of the inspections must be synchronized, so that changes and rates of change can be calculated. In the prior art, the track position measurements are shifted locally to each other. In addition, deviations in the sampling steps can not be ruled out, which is why the track position signals are additionally compressed or stretched relative to each other. Furthermore, the measured values are available at different local points. This is particularly problematic since, during longer measuring runs, the offset as well as the distortion or compression / extension are not constant over the entire length of the measuring run. Thus, a manual, subsequent synchronization is not possible. For the reasons mentioned above, the results of the inspection trips can not be manually synchronized with each other.

For the tracking of changes in the track situation is therefore often resorting to sections statistical variables and these are compared. For example, sections with a length of 250 m are used and the standard deviations are determined for the individual measuring channels - track position parameters. With this approach, the change in the track position quality can be assessed, the temporal change of individual defects - track position deviations - is not possible with this approach.

In order to track the change in individual incident sites, the results of the inspection trips must be precisely synchronized with each other.

It is often attempted to improve the spatial assignment of the measured values to the distance kilometer by means of GPS and DGPS. These improvements are not sufficient for an automatic tracking of the development of the track position and even of individual points of disturbance, since the accuracy of the GPS positioning is not sufficient and the deviations in the scanning steps can not be compensated even with an accurate location. Automatic tracking of individual malfunctions is still not possible.

• laufendes sensorisches Erfassen von Ortsdaten der Gleisgeometrie im überwachten Bereich,
• Auswerten der erfassten Ortsdaten hinsichtlich der Befahrbarkeit und/oder der maximal zulässigen Überfahrgeschwindigkeit des Schienenfahrweges im überwachten Bereich und
• Ausgeben der ausgewerteten Daten sowie Ansteuern von optischen und/oder akustischen Anzeigen und/oder von Regelsystemen.

From the DE 10 2006 043 043 A1 is a method for monitoring rail tracks, especially in the field of temporary support systems for the track body with the method steps:

• continuous sensory acquisition of location data of the track geometry in the monitored area,
• Evaluating the recorded location data with regard to the trafficability and / or the maximum permissible speed of the rail travel in the monitored area and
• Outputting the evaluated data and controlling optical and / or acoustic displays and / or control systems.

It describes a stationary system for monitoring the track position, which is used primarily at construction sites. There is no synchronization of the measured data.

The EP 1 213 202 B2 describes a method for mapping the track condition via signals from various sensors attached to wheels and bogies, which are fed to an evaluation unit and transmitted to a control center, where they are assembled to form an image of the track condition. Again, there is no synchronization of the measured data.

Furthermore, from the US 5,579,013 A1 and the US 6,044,698 A1 Further methods for determining track position errors without synchronization of the measured data known.

In US 2009/0070064 A1 a method is described which can determine a trend from several measurements of the vertical track position deviations. For short sections, less than 100 m, the vertical track deviations are superimposed by means of correlation methods. In this case, it is assumed that the vertical track position deviations, individual errors, change only slightly in the vertical direction and thus a correlation function can be determined. Furthermore, it must be assumed that the local sampling rate is constant and the same for each measurement. If track deviations are eliminated in the course of repair work, this procedure can not be used and trend development is not possible. In addition, an automatic trend calculation is not possible because, despite constant local sampling rates, the measured values of different measurements are located at different locations.

The invention has for its object to synchronize the measurement results of inspection trips each other so that as a result synchronized measurement results are available that allow automatic evaluation and evaluation of the results. Furthermore, there is the possibility that medium and long-term maintenance measures can be derived automatically from the track position measurements and the temporal development of track-bearing faults.

This is inventively achieved by the content of claims 1 to 8.

In the method according to the invention, the results of an inspection run are selected in the first step, which serve as a reference measurement. Since the location-dependent sampling steps .DELTA.x of an inspection measurement are not constant, the measured values for new locally equidistant x-coordinates are calculated. The new location coordinates (x-coordinates) are determined by recalculating the distance between the beginning and the end of the inspection journey - reference measurement - in such a way that equidistant x-coordinates are created. The associated measured values (y-coordinate) can be determined by means of interpolation calculation. Here, linear as well as non-linear interpolation methods can be used. This step is performed for all measurement channels of a measurement run. The locally processed measurement data of the inspection trip serve as a reference measurement. All further inspection runs are synchronized with this reference measurement.

$$\overline{y_{\mathit{ri}}}=\frac{1}{N}{\displaystyle \sum _{n=1}^N}{y}_{\mathit{ri}}\left[n\cdot \mathrm{\Delta }\mathit{x}\right]$$

$$\overline{y_{\mathit{si}}}=\frac{1}{N}{\displaystyle \sum _{n=1}^N}{y}_{\mathit{si}}\left[n\cdot \mathrm{\Delta }\mathit{x}\right]$$

In order to determine the local displacement and distortion of the measurement data from the inspection run, the following steps are processed. The reference measurement and the measurements to be synchronized are subdivided into half-overlapping sections of, for example, 100 m in length. Depending on the application, other section lengths are conceivable. For each of these overlapping sections, the cross-correlation function Φ is calculated. $${\mathrm{\Phi }}_{{\mathit{y}}_{\mathit{ri}}{\mathit{y}}_{\mathit{si}}}\left[k\cdot \mathrm{\Delta }\mathit{x}\right]=\frac{1}{N}{\displaystyle \underset{n=1}{\overset{N}{\Sigma }}}\left(y_{\mathit{ri}}\left[n\cdot \mathrm{\Delta }\mathit{x}\right]-\widetilde{y_{\mathit{ri}}}\right)\cdot \left(y_{\mathit{si}}\left[n\cdot k\right]\cdot \mathrm{\Delta }\mathit{x}-\widetilde{y_{\mathit{si}}}\right)$$

$$\widetilde{y_{\mathit{ri}}}=\frac{1}{N}{\displaystyle \underset{n=1}{\overset{N}{\Sigma }}}{y}_{\mathit{ri}}\left[n\cdot \mathrm{\Delta }\mathit{x}\right]$$

$$\widetilde{y_{\mathit{si}}}=\frac{1}{N}{\displaystyle \underset{n=1}{\overset{N}{\Sigma }}}{y}_{\mathit{si}}\left[n\cdot \mathrm{\Delta }\mathit{x}\right]$$

The location of the maximum of the calculated cross-correlation function Φ indicates the local displacement of the two sections y _ri and y _si to each other. This local displacement is determined for each overlapping section pair. From the local displacements of all sections y _si to the respective section y _{ri of} the reference measurement, the km offset is calculated by means of interpolation method. Since the mileage offset is generally not linear, it will be automatic as well as local displacement the distortion or strain and compression, which are non-linear over the entire measurement. The thus determined km offset is used for all measuring channels of the measurements to be synchronized.

Finally, for each measuring channel, the measuring points are calculated at the points of the reference measurement by means of interpolation methods. Here, linear as well as non-linear interpolation methods can be used.

The inspection measurement was synchronized with the reference measurement. The individual measuring points are in relation to the x-coordinate, abzufastpunktgenau one above the other. An automatic evaluation and evaluation is now possible.

For the reference measurement and the measurements to be synchronized, a local scan of 0.10 m is selected.

Embodiment:

The invention will be explained below with reference to an exemplary embodiment.

• die Längshöhen der linken und rechten Schiene - z_links und z_rechts,
• die Richtungsabweichungen der linken und rechten Schiene - y_links und y_rechts.
• die Spurweite - sp,
• die Krümmung - kr,
• die gegenseitige Höhenlage - gh.

As an example, two measurements from inspection trips to the track inspection at the Deutsche Bahn AG serve. During these inspection trips, the following track-laying parameters are measured:

• the longitudinal heights of the left and right rails - z _left and z _right ,
• the directional deviations of the left and right rails - y _left and y _right .
• the gauge - sp,
• the curvature - kr,
• the mutual altitude - gh.

The used track system is in illustration 1 shown.

The synchronization takes place in such a way that any number of additional measuring channels can be synchronized. If the reference measurement has been prepared as described below, any number of inspection measurements can be taken with the reference measurement synchronize. To clarify the procedure, an inspection measurement is synchronized with the reference measurement in the application example.

The FIG. 2 shows a section with 100m track length. The local shift of the measurement 1 with respect to the measurement 2 can be clearly seen. This is about 30 m.

In the first step, an inspection measurement is prepared in such a way that the measured values can be assigned to defined distance kilometers and that a constant scanning step of, for example, 0.10 m results, other scanning steps are also possible. This processing is done by evenly dividing the section between the first and the last measuring point in equidistant increments. Since the new location coordinates thus determined do not correspond to the original ones, the measured values must be calculated at the new location coordinates. This is done in the exemplary embodiment by means of interpolation with cubic splines. In general, linear interpolation methods can also be used. For all measuring channels of the reference measurement, the corresponding measured values are recalculated to the method described above.

As a result of the first step all measuring points of all channels of the reference measurement are available on fixed defined route kilometers (x-coordinates). Deviations of the scanning steps were also corrected in the first step.

In the second step, the further inspection measurements are synchronized to the reference measurement. This is done in such a way that in each case one measurement channel from the reference measurement and the corresponding channel are selected from the inspection measurement to be synchronized. In the application example, the track width was selected because the track width changes much more slowly than, for example, the longitudinal heights or the deviations in direction compared to the other track position parameters. The entire measurement is subdivided into sections of 100 m in length, whereby the individual sections overlap each half, ie 50 m. For each section, with the track of the reference measurement and the corresponding section of the inspection measurement to be synchronized, the cross-correlation function Φ is calculated and normalized.

FIG. 3 shows the cross-correlation function for a section.

The position of the maximum of the cross-correlation function indicates the displacement of the selected sections relative to each other. In the exemplary embodiment, the displacement is 29,6 m. For each section pair, the mutual displacement is calculated in this way. From the shifts of all sections of the entire inspection journey, the offset of the kilometer is determined. Since the shift of the individual sections can be different, the kilometer correction is determined by means of nonlinear interpolation (spline interpolation). Since the displacement of the individual sections is different, the distortion or extension and compression of the signals is automatically determined in this way.

FIG. 4 shows the calculated kilometer offset.

The kilometer offset thus determined is applied to all measuring channels, the measurement to be synchronized.

In the last step, the measurement points of all measurement channels of the measurement to be synchronized are calculated at the mileage points of the reference measurement. This can be done with interpolation. In the application example, spline interpolation was used.

The result is synchronized measurements in which the sampling steps have been corrected and the measurement points of all measurement channels of all measurements are at the same km coordinates.

Automated further processing is now possible.

FIG. 5 shows the synchronized measurements.

Key to symbols / symbols:

Δ x

Ortsinkrement; locally equidistant sampling interval

Φ _{y ri y si}

Cross-correlation function of a section y _{ri of} the reference measurement and the corresponding section y _{si of} the measurement to be synchronized

y _ri

Measured values of section i of the reference measurement y _r

y _si

Measured values of section i of the measurement to be synchronized y _s

k

Index of cross-correlation function; Shifting the sections y _ri and y _si to each other

n

Index of the measured values

y _ri

arithmetic mean of the measured values of section i of the reference measurement y _r

y _si

Arithmetic mean of the measured values of the section i of the measurement to be synchronized y _s

i

section number

N

Number of measuring points - samples - in a section i

z _{on the left} , z _{on the right}

Longitudinal height of the left and right rail

y _left, y _right

Directional deviation of the left and right rail

sp

gauge

kr

curvature

gh

Mutual altitude

Claims (11)

Method for automatically synchronizing measurement data, in particular track geometry data, wherein the measurement results of different measurements, in the case of track geometry measurements of different inspection journeys, are spatially shifted and / or distorted relative to one another and the local displacement and / or distortion does not occur constantly over the entire measurement,
characterized in that • a measurement is selected as a reference and a local scan is calculated to be constant for the entire measurement, and • the measured values of all measuring channels for the newly calculated local sampling values are determined by means of interpolation methods and • that this measurement is used as a reference measurement and the reference measurement and the measurement to be synchronized are divided into sections, the sections overlapping and for each section pair the local displacement is determined by a maximum of the cross-correlation function of a measurement channel of the two measurements, and • an offset, which is generally not linear, is determined from the local displacements of all sections and used for all measurement channels of the measurement to be synchronized; and • thus the displacement and / or distortion are compensated and • the measured values of the measurement to be synchronized are calculated at the locations of the reference measurement and • Thus the measured values of the reference measurement and all the measurements to be synchronized have the same location coordinates. A method according to claim 1, characterized in that a local scan of 0.10 m is selected for the reference measurement and the measurements to be synchronized. A method according to claim 1, characterized in that the reference measurement and the measurement to be synchronized are divided into sections of 100 m in length. A method according to claim 1, characterized in that the sections for the determination of the cross-correlation function each overlap in half. A method according to claim 1 to 4, characterized in that the measurements of the track geometry of railroad tracks are synchronized with each other. A method according to claim 1 to 5, characterized in that the cross-correlation functions of the individual sections are determined for the gauges. A method according to claim 1 to 6, characterized in that is determined as an offset, calculated by the method, relative local displacement, compression and / or extension of the synchronized measurement for reference measurement. A method according to claim 1 to 7, characterized in that the section lengths are chosen so as to give a possible good correlation between the reference measurement and the measurement to be synchronized and that the section lengths are longer than the offset of the reference measurement to be synchronized measurement. A method according to claim 1 to 8, characterized in that the reference measurement is selected so that the cross-correlation with a measurement to be synchronized results in the largest possible cross-correlation coefficient or that the oldest or most recent measurement serves as a reference measurement. A method according to claim 1 to 9, characterized in that the constant scan is calculated such that a locally equidistant increment for the entire inspection measurement is formed. A method according to claim 1 to 10, characterized in that the determined offset is a measure of the local offset of the sections to each other.

Images