WO2008052643A2 - Method and device for evaluation of measurement data in railway track circuits - Google Patents

Abstract

A device and a method for operation of railway track circuits are disclosed, which have an associated section length which is restricted at both ends by track isolation. The following procedure is used for reliable identification of the states of FREE, BUSY and DISTURBED within a predetermined time limit. A transmitted signal is fed into the rail body at one of the two section ends. The received signal is output at the opposite end, and is passed on for analysis. The transmitted signal that is fed in is sinusoidal. The frequency is switched cyclically between two discrete values. Three evaluation channels are operated at the receiving end. The information is A/D-converted for all three channels. Two channels have a bandpass input filter, which is matched to pass the channel (1) with the frequency that is currently being transmitted, for the other channel (2) with the frequency which is not being transmitted. The busy state is detected by the channel (1). The currently occurring disturbance influence is measured by the channel (2). The channel (3) carries out sampling in the respective transmission interval, and evaluates the calculated FFT results. Further data is obtained from this information. Data that may be mentioned in this case includes isolation shock bridging, disturbance influences and further process-relevant variables.

Description

Method and device for evaluating measured data in railroad track circuits

The present invention relates to a method and apparatus for evaluating measurement data in railway track circuits. Further, the invention relates to a method and apparatus for track vacancy using track circuits.

For the management of railways, the information about the occupancy of a track section by a vehicle is absolutely necessary. It plays an important role in the setting and dissolution of routes. There are many different techniques for recognizing occupancy. The starting point of this invention is the widely used track circuit with the classical technical approach of the circuit. The functional principle is relatively simple and will be briefly introduced. A rail of the track to be monitored is separated and insulated at two points. The transmitter applies voltage to one end of the isolated section. The receiver at the other end of the section evaluates this voltage. When a train enters the section, its axles short the two mutually insulated rails of the track and lay the track to the railroad ground. The received signal is thereby suppressed. The evaluation after the receiver reports the section as occupied.

Building on this basic principle, there are a large number of very different products. They differ considerably in terms of the selected transmit signal and the evaluation in the receiver. However, there are also so-called shock-free track circuits. However, these have some overlap at the boundaries between two neighboring circles. As a result, selective assignments are not possible, for example in the area of turnouts and crossings.

One embodiment known in the art is the classic DC track circuit. This principle operates a DC Circuit at the track section and therein are serially the two rails with the variable size of the ballast resistor. The axle shunt during a drive reduces the resistance in a circle. The principle works as shown in FIG. The simplest solution is to use a relay as a receiver serially in the circuit for detecting the occupancy of the section.

Another known embodiment provides for the supply of the GSK (track circuit) by means of a three-phase network. It comes in a driving to a detuning a device of a kind of differential transformer or a kind of electric shaft or motor. At the manufacturer Siemens, such a product is referred to as a motor relay. When driving, there is a change in the circle with respect to the following parameters, namely frequency spectrum, phase, power. The mass of the motor rotor acts as an inertia filter with a pick-up delay to suppress short-term interference.

The principle of a GSK operated with a voltage over 100V AC (usually 230V) with 50 Hz or 60 Hz, in continuous operation or in pulsed operation. The comparatively high voltage is used for breaking up the insulating layers, e.g. Rust used on the rail rolling surfaces. Due to the high performance, the required fault clearances are created.

Based on this state of the art, it nevertheless remains desirable to be able to further improve the reliability of track circuits and their robustness against electromagnetic interference.

In the course of the present invention this object is achieved by a system and a method for track release of a section of a track section, in which: a) a length of the section is defined by a rail track interrupted at two ends; b) at one end of the rail line an alternating voltage with two alternating frequencies is fed as an input signal; c) an output signal is tapped off at the other end of the rail track; d) the tapped output signal is analyzed for its components in the two frequencies; and e) it is decided in dependence on the analysis by means of limit comparisons which state the track section has.

In this way, to determine a track occupancy, the defined AC voltage fed with a defined pulse duty factor is used, which preferably has no frequency matching with the harmonics to the frequencies of the traction supply used. The selected sampling frequency may be set in response to the required response times for providing the fuse-technical operating conditions.

The method can be further improved in one embodiment of the invention in that the very powerful signal processing based on mathematical calculation of processes is used. Processing takes place in real time with a signal-data acquisition grid suitable for the task. The digitized values are specifically fed to the evaluation channels designed for the respective task.

Preferred embodiments of the present invention will be explained in more detail with reference to a drawing. Showing:

Figure 1 shows a schematic representation of the basic principle of a track circuit;

Figure 2 shows a received and filtered signal with interval division;

Figure 3 threshold and a schematic level profile for a track occupancy; Figure 4 in a schematic representation of the emergence of maximum interference voltages by rail currents;

Figure 5 transmission frequencies for the track circuit and harmonics of the traction currents for a frequency of 16 2/3 Hz;

FIG. 6 shows an FFT of the previous and the adapted transmission frequencies;

FIG. 7 is a schematic representation of the functional blocks of the evaluation device;

FIG. 8 shows an IIR receive filter;

Figure 9 shows the effect of coefficient change on the output signal;

Figure 10 Filter characteristic at 1600 Hz (left) and 1724 Hz (right);

Figure 11 is a schematic representation of the evaluation of the digitized received signal;

FIG. 12 shows the functional blocks in a third receiving channel;

FIG. 13 shows a state graph of a general free-field system;

FIG. 14 shows a sinusoidal interference signal;

FIG. 15 noise as interference signal;

FIG. 16 shows a Gaussian pulse as a short-term interference;

FIG. 17 shows the simulation of a free undisturbed section;

FIG. 18 shows the course of the coefficients at the selected observation times (sampling times);

Figure 19 shows the time course of the coefficients in a regular occupancy;

Figure 20 shows the influence of an adjacent track circuit in a Isolierstossfehler;

FIG. 21 the influence of a sinusoidal disturbance;

FIG. 22 shows the influence of a further sinusoidal disturbance; and

FIG. 23 shows the influence of a further sinusoidal disturbance.

The following describes the universal track circuit (UGSK) of Siemens Schweiz AG. The system consists of two parts: the part "outdoor installation", which establishes the physical connection to the infrastructure, ie the track section, and the part "indoor installation" in the signal box with the associated electronics for signal generation and evaluation of the free resp. Busy condition.

The outdoor unit essentially consists of the power transformers with wiring at both ends of the section transforming the transmitted high voltage on the cable to the interlocking into a track voltage of a few volts and equipped with a high-pass filter to protect the electronics of the transmitter and receiver against the high-energy interference effects of the traction currents from the 16.7 Hz or 50 Hz traction networks of the railways.

The part "indoor unit" consists of a strictly two-channel version of the power supply, a microcontroller for the monitoring of the functions as well as a DSP for the signal processing as well as the user interfaces A potential separation of the transmitting and receiving line is realized The transmitter and receiver are physically located next to each other for reasons of noise control.

The UGSK uses a pause-modulated sinusoidal signal with a selectable fundamental frequency of 137.5 Hz, 175 Hz or 225 Hz. The system must always be set so that the transmitter never uses the same frequency as that of the neighboring section. The level of the transmitter is adjustable and can be adapted to the external conditions. The ratio between transmission phase and transmission pause is 3: 2 with a period of 200 ms.

The receiver has a 99th order digital FIR filter at the input. Figure 2 shows the timing of the transmission signal and the received signal after the filter in the undisturbed state. To determine the occupancy state, the transmitter cycle is subdivided into five intervals, each with a T = 40 ms duration. The burst intervals 2 and 3 are used to detect an occupancy. The gap interval 5 allows the detection of disturbances due to currents in the rail. The two ramp intervals 1 and 4 are in the Evaluation not taken into account, since their information content is not usable.

The total division of intervals 1 to 5 is compared to the cycle of the transmitter due to the lead times in the track circuit shifted by 11 ms. The shift better places the gap interval over the signal pause.

For evaluating the route condition, the amount of the received signal is integrated over an interval.

formula 1

FIG. 3 schematically shows the threshold values which serve to evaluate the calculated level P. When the level of a burst interval reaches a threshold, this interval is considered free. If a certain number of burst intervals are detected as free, the track section is considered free. If the level of some intervals falls below a lower limit, the section is occupied.

The evaluation of the gap intervals is used for fault detection. If an inadmissibly high level is detected within a gap, the interval is considered disturbed. The use of a pause modulated signal has indisputable advantages in detecting interference, especially in the case of perturbations in the passband of the filter. Nevertheless, the selectivity of the filter must be compromised because the on / off-swing phases can not be used for evaluation.

A busy message must be delivered according to Table 1 after 300 ms. This requirement limits the order of the filter and thus the Störimmunität. In order to illustrate this problem more precisely, the following section describes the various faults. They also allow a view of the limits of new procedures. Another method also consists of the two main functions sender and receiver. The transmitter also sends a signal to the track section, u = f (t). However, two or more frequencies are advantageously transmitted alternately or simultaneously from sinusoidal sources. Other functions are conceivable. On the receiver side, the signal influenced by the transmission characteristics of the track section and the respective voyage is received. The received signal is now fed to an A / D converter, which converts the input signal into digital values at a relatively high sampling rate. A defined window with a number of sample values (2 ⁿ ) obtained from the AC / DC oversampling rate reduced discrete measurement values by a method such as averaging, 1 from N, peak value, minimum value (and others) shifts "real time At each step, the result is fed in the form of discrete amplitude values from the monitored frequency range to the available evaluation units, which are then able to evaluate the temporal course of the amplitude and / or frequency according to their predefined criteria specific decision such as for occupancy.

The disturbances which can occur in railway operation are divided into static and dynamic disturbances. Static disturbances include all extraordinary operating cases that are related to faults in the infrastructure (track system and safety device) and are not necessarily triggered by a vehicle. The disturbances can occur suddenly, but remain for a long time. They usually require an intervention of the staff. The dynamic disturbances, on the other hand, result from the energy consumption and acceptance of the vehicles with the accompanying electrical and electromagnetic influences in the regular railway operation.

An exact description of the dynamic disturbances is difficult because they are very diverse and duration, frequency and amplitude of a large variability are exposed. The previous system distinguishes three different static disturbances. a) The ballast disorder

For the operation of the track circuit is considered as a critical fault, the ballast disorder. It is detected when the receive level is between the occupancy threshold and the idle state. The system can no longer determine with certainty whether the section is free or busy. The reason for this may be that the modulus of conduction has risen impermissibly, so that the level without vehicle in the track is too much damped. On the other hand, the axle shunt resistance caused by rust-applied wheel treads or rail running surfaces can be so great that the axle shunt and thus the level are not sufficiently small despite their occupancy.

The fault disappears when either the vehicle has left the section or the infrastructure has been repaired.

b) Override

Another critical disorder is the override. It occurs when the transmitter level is set too high. The receiver can in this case also make no certain statement about the occupancy state of the track, since the level does not fall below the threshold for poor occupancy. This error can also be corrected only by user intervention on the transmitter.

c) insulating shock bypass

An isolating shock bypass means that two adjacent track sections are no longer separated. The track circuit will detect the transmitter of the neighboring section in the unused frequency band and thus detect the fault. This disruption requires intervention of the staff in the infrastructure. These disturbances can be detected solely by evaluating the burst intervals by means of threshold values. The consideration of the gap interval is not necessary. The dynamic disturbance critical to a track circuit is traction current disturbance. The traction current flows via the overhead line, through the vehicle and then back to the feed. Part of the stream flows back over the rail, another part through the earth cable to the catenary masts and a small part through the soil. In multi-track systems, the return current can also flow back through the neighboring tracks and cause disturbances there without the neighboring track being traveled. A railway system thus acts as a multi-conductor system for the return flow.

For the worst-case scenario it is assumed that one vehicle is outside the section but the infeed is on the opposite side.

The entire traction current should only flow back through the rail, as the existence of a ground wire and a consistent grounding is not guaranteed.

The voltage at the receiver results in:

u _e = u _s + - ^ - ■ I _n . i (R '+ iωL') ■ I

Formula 2

It can be seen that the disturbance is primarily a function of the disturbance current amplitude _Tr and the section length 1. The interference voltage is divided according to the impedances on transmitter and receiver. The resistance and inductance coating from the equivalent circuit diagram in Figure 4 must be halved since these values apply to both rails together.

The voltage drop along the insulated rail is neglected because its series resistance is much smaller than the terminations. The simple multiplication with the section length is permissible since there are no leakage effects in the grounded rail. The modern three-phase asynchronous drive technology has in the last 20 years the series engines and the chopper technology as well DC drives in railways largely displaced. The advances in power electronics enabled ever higher vehicle performance and sensitive control. However, converter vehicles have a considerably greater influence on the track facilities. The coupling takes place either via electromagnetic fields or galvanically through the traction currents. Track circuits, unlike axle counting systems, are less susceptible to electromagnetic interference than to the direct disturbances due to traction currents described above.

The main change through the use of converter vehicles is that the frequency components in the traction current have become almost arbitrary. The largest disturbances still occur due to the fundamental frequency of 16.7 Hz or 50 Hz and the corresponding, damped harmonics. Added to this are the disturbances that couple in from the drive to the primary side of the traction vehicle transformer. Its frequency is directly related to the current fundamental frequency of the motor and thus its speed together. In addition, the return conductor of the Zugsammeischiene (1000V supply) also leads through the rail.

Due to the large variety and variety of connected consumers (including inverters), the interference is difficult to detect. The disturbance can consequently occur in a broad frequency band and thus also in the passband of the filter in the track circuit.

The disturbance in the passband of the filter is a real threat to the integrity of the occupancy statement. On the one hand, the transmission signal can be canceled and a wrong assignment can be generated. On the other hand, the level can be so large despite actual occupancy at the receiver, that a false free message arises. Consequently, the receiver can not make a reliable statement about the occupancy state under this disturbance. The fault can only be detected by evaluating additional information. In the solution explained above, the level of the gap interval was evaluated for this purpose. A fault is detected if the level exceeds a threshold (install disturbance current limits). The facts described above push the UGSK system described above to its limits. Again and again, an accumulation of effects by the traction currents can lead to interference in the evaluation of the received signal. As a result, the UGSK is blocked.

An incorrectly recognized occupancy puts the system in a safe state, but a required intervention of the personnel represents a safety-critical process and thus generates delay minutes which prevent operation.

The aim of the object / invention is therefore to achieve increased immunity to traction influences. This improves availability and reduces the likelihood of safety-critical interventions, or it can increase the allowable intercept length while maintaining consistent availability and security.

The most important preliminary considerations and boundary conditions are discussed below. The characteristic of the receiver is a central point for the improvement of the interference immunity. The transmission signal contains no information in the sense of communications engineering. The receiver need not be able to decode and transmit information from a broadcast channel. Rather, in the received signal a fault must be reliably detected and reasonably evaluated. The evaluation criteria are based on the required minimum security and on a high availability of the system.

Railway operators define permissible levels and usable frequencies for the use of track circuits. These are based on railway measurements on routes. For new track circuits, the useful frequencies should not be less than 200 Hz. Due to a sensible reuse of the old outdoor facilities, frequencies above 250 Hz are not suitable. FIG. 5 shows a further aspect for the selection of the transmission frequencies. These are advantageous between the harmonics of the fundamental frequency. Building on these considerations, the frequencies fl = 208, 1 Hz, f2 = 224.2 Hz and f3 = 241.3 Hz were set. All other comments regarding the new methods will use these frequencies. Three frequencies are necessary in order to be able to distinguish the track circuits from each other in switch fields. This allows the detection of the fault of Isolierstossüberbrückungen.

With the frequency, the amplitude of the transmission signal is largely fixed. It has to be big enough to ensure reliable detection in poor bedding. The other limit is given up, so as not to overload the recipient with good bedding and thus small losses. However, a modulation of amplitude and frequency is conceivable. In addition, time limits are imposed on the track circuits with regard to recognition of the occupancy state. The evaluation must reliably recognize the change of an occupancy status from the time of admission within the specified time limit. The operator defines this time, which may pass from the occurrence of the event to the message to the signal box. The detection must therefore be such that the output units, such as e.g. the safety relays can be controlled reliably and read out again. The specified 50 ms are sufficient for experience.

Table 1: Timing requirements for a track circuit

The occupancy of a track section by a short circuit with a metallic construction part between the two rails has the consequence that the information about an occupancy only in the ampere Litude of the received signal is included. The assessment of the occupancy state is only effective by comparing the signal size at the receiver with fixed threshold values. There are no known methods that allow selective / adaptive provision of thresholds. Excluded from this statement are adjustments of the sleepers, which are derived from climatic parameters or also from parameters of the conductivity of the track bed. It is also possible to adjust the transmission level accordingly.

Definition of the required thresholds for the intended function

N name designation

1 pl threshold for free message

2 P2 threshold for occupancy message

3 Pmax threshold for overdrive

4 piso threshold for shock bridging

5 ptr threshold for traction flow

Table 2: Level Rating Thresholds

The threshold p _max is the highest threshold. The condition Pl> P2 defines the forbidden intermediate range between the free message FM and the busy message BM.

In the preceding paragraphs the known and applied principles 1-4 have been described for GSK with their general characteristics. In this case, the USGK has been described in detail with its impact environment on the basis of the example described above.

As a result, now the new method is shown, which has significant advantages over the principles described so far and can provide a much higher diversity evaluation performance than the above principle. There is a much extended process with new properties. The subdivision of the principle track circuit in the individual function blocks is shown below. The stations in the signal path are suitable for explaining the entire new range of functions.

The track circuit electronics include the circuit

A signal transmitter, which feeds a transmission signal (frequency and amplitude) suitable for the network of the track and the traction influences at one end of the section via the launching network, and the

Signal receiver, which is adapted to receive correctly at the other end of the section via the Auskoppelnetzwerk from the track section, the exiting by the respective state (free or occupied) of the track and by the traction effect influenced transmission signal, and the

Signal evaluation, which performs the data processing by an optimized method. These results are then fed in the data density per time unit which is expedient for the respective task to the processing units which are advantageously optimized for each task. This continuous evaluation of the received signal allows a safe detection of the conditions FREI, BELEGT and STORED and beyond the detection of other influencing variables. and the

Output unit for the output of the FREI, BELEGT or ERROR message to the signal box, by means of safety relays for easy read back of the output status.

The transmitter

The transmission signal and the selective evaluation methods are matched to one another. The method consists in the generation of a quasi-continuous transmission signal, which nevertheless allows a reliable detection of interference, it is described below. The transmitter uses, in contrast to the previous methods, two frequencies, which are alternately output in a symmetrical grid. These two frequencies are determined from usually three or more frequencies determined by the infrastructure manager.

By means of the combinatorial "2 of 2" the demarcation to adjacent track circuits is achieved and the recognition of the fault of a track isolation bridging is recognized when an unshifted frequency occurs with the resulting physical characteristics.The choice of frequencies has in the proposed solution to the Method itself no influence, a corresponding parameterization of the transmitter and processing algorithms are required.

The detection of a valid assignment within 250 ms is optimally supported by the chosen solution with the two transmission frequencies. The signaling structure of an occupancy must therefore be able to be recorded continuously. Although the use of a continuous signal offers such a possibility, it has disadvantages in the detection of acting disturbances near the transmission frequency.

To avoid these disadvantages, the transmission signal is parameterized as follows.

The transmission duration for the two frequencies (each channel) is set to 170 ms by way of example.

The transmission frequencies are adjusted accordingly with regard to the signal evaluation method, that is, they may be slightly shifted to the specifications of the railway operators.

For the new solution, the three predefined frequencies (two are used in a section) are shifted into the discrete grid of the Fourier transformation. An adaptation in compliance with all requirements on the side of reception and evaluation in this case leads to the following frequencies. This representation is exemplary, it is possible to make any determinations within the framework of the FFT properties; fl = 206.25 Hz, f2 = 225 Hz and f3 = 243.75 Hz. The following FIG. 6 clarifies the advantages of the adapted definition.

The recipient

There are no special requirements for the recipient. Rather, he has the task of level adjustment for the purpose of protecting the inputs from overdriving. In addition, an adaptation to the line and the Empfangsübertragers to achieve a minimum impairment is required.

The evaluation

The transmitted from the section taken by the respective occupancy state and by the traction current transmitted signal is fed to an A / D converter.

The conversion works fast and over-sampling at a rate of about 10: 1 according to the characteristics of the input signal. The method used combines the approaches of various methods to produce a quasi-continuous transmission signal, which nevertheless allows a reliable detection of interference. In contrast to the previous methods, the transmitter uses two frequencies, which are used alternately. The currently transmitted and received signal in each case one of the two frequencies associated with the system is able to represent the current occupancy state or its time history. On the currently not sent frequency it is possible to evaluate the incident disturbances. Furthermore, the failure of track isolations between the sections may result in non-GSK frequencies of the adjacent circuits being detected. The difference between the methods used to date is that filters are not switched on and off, but only just accessed and used in a parameterized set to the current transmission frequency in a timed interval in the evaluation algorithm.

For the evaluation, the received signal is generally discretized by means of an A / D converter and the current signal course in one Time window analyzed by means of an FFT. The duration of the window is matched to the frequency spectrum that is important for occupancy detection. In the evaluation, two methods are used, the continuous method of evaluation and the discontinuous. If a short decision time is required for the reliable detection of a state such as that of the section occupancy, the continuous evaluation is applied. For the release of the section as well as for the detection of a Isolierstossüberbrückung and also for the monitoring of the traction interference, the discontinuous evaluation is advantageous. The discontinuous method allows evaluation of the frequency spectrum in any number n of resolution steps from the continuous process. An advantageous definition is the analysis of the waveform of a time window with a defined time position for the transmission frequency switching.

Details for continuous evaluation

A 256-point FFT window advances step by step with each arrival of a digital input signal value, adds the currently-converted instantaneous peak value, and eliminates the oldest value (FILO). After each forward step, an FFT is performed. With respect to the currently evaluated window, the two frequency components occur in variable proportions, the sum of the two components being constant. As a result, the periodic change of the transmission frequency without any effect on the processing, the continuity is maintained.

Details for the discontinuous evaluation

The FFT window of the continuous evaluation is also used unchanged for the discontinuous evaluation as a data source. However, data is not taken in each step, but after N steps or at a defined time relative to the transmission interval for the two different frequencies. This evaluation over time allows the detailed detection of all events that have no synchronicity with the system timing and temporally incurred in relation to the system timing only short-term (transient). Evaluation Algorithm 1 Section occupancy (continuous) The amplitude value of the currently transmitted frequency is taken continuously from the frequency components calculated by FFT from each generated FFT window and evaluated according to the allocation information. The two alternately transmitted frequencies have no influence on the evaluation method. The time profile of the amplitude of the transmission signal is determined by means of an amplitude evaluation for the recognition of the states 1 and 2 listed in Table 1, the occupancy and the free message), taking into account the time criteria defined in Table 2 (see also FIG. 19).

Evaluation algorithm 2 traction disturbance (discontinuous)

The analysis of the received amplitudes in the currently not sent frequency identically Algorithm 1 allows an assessment of the disturbing effects of this frequency at the current time. Since such disruptive effects due to traction are usually significantly greater than the pause time of the sampling raster in terms of time, there is also a requirement for the evaluation of the influence of the transmission signal.

Evaluation Algorithm 3 Further influences (discontinuous)

In addition to the evaluation of the course of the amplitude over the time of the two transmission frequencies and the thus possible occupancy statement, the penetration of working frequencies from neighboring zones can also be detected as an error when an isolating shock bypass occurs. The disclosure of such an error is not time critical and can be reported further.

Representation of the method and the device for the detection of occupancy states in a track section is shown in the following figure. Further evaluations by means of algorithm 4 to n are possible if required.

The shift of the amplitude response of a digital filter can be done in two ways. For one thing, the characteristic of constant sampling frequency can be adjusted by changing the filter coefficients during operation. On the other hand, the sampling frequency can be changed without modifying the coefficients.

This method offers the following advantages:

The quasi-continuous transmission signal enables the same selective filters as are possible with only one transmission frequency. The immunity against interference by the reduction of the bandwidth is therefore maximum. The detection of disturbances is permanently possible. Neither for occupancy detection nor for fault detection therefore worst-case cases must be assumed for the occupancy time. The safety in the detection of a noise voltage is comparable or better than the described UGSK system. Simulations show that the use of two filters is not without problems. All switching operations during operation lead to reactions in the output signal of the filter. In addition, the complexity in the field of A / D conversion is increased, since at worst two separate transducers must be used for one signal. Function block A / D conversion

In the event that the adaptation of the filters is achieved only by changing the coefficients, the transducer system can be designed as follows. The oversampling frequency is 12.8 kHz and the useful sampling frequency is 1600 Hz.

In all other cases, additional frequencies must be determined. The ratio of Nutzabtastfrequenz and oversampling should be the factor 8. In the intervals in which the transmitter transmits at the lower frequency, the previous sampling rates are used. The payload frequencies for the higher frequency intervals are calculated in Table 3. The resulting shift of the digital anti-alias filter can be tolerated in this range.

Transmission frequency Phase 1 Phase 2

208.1 and 224.2 Hz 1600 Hz 1724 Hz

208.1 and 241.3 Hz 1600 Hz 1855 Hz 224, 2 and 241, 3 Hz 1600 Hz 1722 Hz

Table 3: Sampling frequencies for shifting the filters

In the implementation, the necessary oversampling frequencies can be obtained from the high system clock of the DSP.

The converter architecture with anti-alias system described in Method I can be adopted unchanged.

The purpose of the filter 1 is to isolate the transmission signal as well as possible, so that only its amplitude is evaluated. Such a filter has been designed for a continuous transmission signal. This filter can be adopted in this method, since the time requirements are identical when using the quasi-continuous transmission signal. To take advantage of this, the filter has to be adapted to the transmission frequency in each interval. The filter.

IIR filter design: The receive filter can alternatively be designed as a feedback system. The filter order is much smaller than in a FIR system. In addition, the filter order has little effect on the delay of the filter. This is mainly determined by the required width of the transition area. The blocking attenuation is determined analogously with - 4OdB. The desired smooth amplitude response in the pass band is achieved by the design as Chebyshev II filter. The result of the design is shown for each transmit frequency in FIG.

The IIR filters have a -6dB bandwidth of 12 Hz. Two approaches are conceivable for this.

The amplitude response of the filter is shifted by replacing the coefficients in operation on the frequency axis. This technique is known by adaptive filters. For this type of filter, an algorithm is designed which changes the coefficients of a filter online to realize a desired behavior. Such Systems are used, for example, for system identification or echo cancellation.

In this method, however, the filter should not be modified continuously, but only at the known times at which the transmitter changes the frequency. The advantage of this method is that no changes must be made in the scanning system and the sampling rate always remains constant. The disadvantage of this technique, however, becomes clear in FIG.

From the 1000th sample, the input signal was switched from 208.1 Hz to 224.2 Hz and the filter was loaded with the corresponding coefficients. The filter shows a transient behavior after the transition, which renders the evaluation of the amplitude useless. In addition, it has to be considered that adaptive solutions always require a costly safety case if security requirements are met.

The adaptation to the signal frequency is already carried out in the function block A / D conversion. For this purpose, the sampling rate must be raised or lowered accordingly. The filter, which, like any digital system, is referenced to the sampling frequency will not detect a change in frequency in the corresponding sampled input signal and will consequently produce an output signal as in the continuous transmit signal method. The designed filters can be taken over.

Function block filter 2

The purpose of the filter 2 is to isolate interferences in the range of the transmission frequencies and make them assessable. In this method, the filter is always set to the currently unused transmit frequency. Since the transmission frequency changes periodically, interference on both frequencies used can be detected. The information obtained corresponds to that of the gap interval in the previous system UGSK. The problem with switching filter 2 is that the direction of the shift on the frequency axis is exactly the opposite of that of filter 1. This relationship is shown in FIG. When switching to the higher sampling frequency, both filters are shifted in the direction of higher frequencies. However, the filter 2 should be tuned to the lower frequency. A first approach is to use a different sampling rate for both frequencies. However, it follows that at least the entire digital part of the A / D conversion must be dual-channel. However, on closer examination of the requirements for fault detection, simplifications can be assumed. For example, it is not absolutely necessary for the noise level to be continuously monitored. It is sufficient if an impermissible interference voltage can be detected at the end of an interval. The filter 2 can thus be reset every time it is switched. The system is then allowed to settle and the amplitude is evaluated. This approach requires a new determination of the length of the transmission phase with a frequency.

For the filter 1, the length of the transmission intervals until switching does not matter because the change is compensated by the sampling. If the same order as for filter 1 is appropriately set for filter 2, a settling time of about 170 ms results. For the subsequent evaluation again two intervals of 40 ms each can be assumed. Each frequency can therefore be examined for a fault within 250 ms. The requirement that a fault message can be issued after approx. 500 ms is almost fulfilled.

Since there is a quasi-continuous signal behind the filters, regardless of the strategy chosen, the evaluation of the amplitude can be performed as follows. To detect the occupancy state of a section, an evaluation of the amplitude of the filtered signal is sufficient. The summation of the magnitude of the signal over a given interval is easy to implement. The length of the viewing interval must correspond to at least the period of the transmission signal, but should include several vibrations for safety reasons. The interval length of the existing system of 40ms is adopted. An appendix However, adaptation is conceivable on the basis of new findings or measurement results. The result of the summation is shown in FIG. Each shift of the window gives a new point. It can be seen that the points are subject to a slight vibration. The narrower the viewing window is chosen, the more pronounced the vibration. For this reason, the window must not be too small. From 32 samples the results are satisfactory.

Method III pursues an approach which promises the theoretically maximum immunity and can nevertheless detect disturbances as reliably as the previous system by evaluating a gap interval. Instead of the transmission pause of the existing system is transmitted in this method on another frequency and tracked a filter in the receiver of the transmission frequency. Another filter is tuned to the unused frequency and allows the evaluation of interference signals. From the various approaches presented and discussed above, the use of a variable sampling rate and a non-continuous filter 2 promises the greatest success. It avoids the need for a second A / D converter and avoids the problems of online tracking of the coefficients of filter 1. The transmission duration of a frequency is set to 250 ms.

The most important operating parameters are:

Parameter value

Oversampling frequency = 12.8

Sampling frequency Is = = 1600 Hz

Transmitting frequency 1 fl = 208.1 Hz

Transmission frequency 2 f2 = 224.2 Hz

Transmitting frequency 3 f3 = 241.3 Hz

Transmission phase channel A ta = 210 ms

Transmission phase channel B tb 210 ms

Table 4: Process parameters The spectrum analysis method uses, in contrast to all other methods presented, additional information in the frequency domain of the received signal. The signal processors and fast transformation algorithms available today make a fundamental analysis of the received signals seem sensible. The receiver of this method has a simple structure consisting of three functional blocks (FIG. 12). A discrete transformation is applied to the received digitized signal. The calculated coefficients contain all the necessary information to be able to evaluate the occupancy status and fault of the section. For example, short-time discrete Fourier transformation and wavelet transformation are used as transformations. The definition of the parameters of the STFT has an effect on the transmission signal to be selected. This mechanism is described below.

The converter architecture with anti-alias system described in Method I can be adopted unchanged.

For the evaluation of the received signal two transformations come into consideration, which are described in the following sections. First, the occupancy detection is realized with the help of the STFT. It uses fixed-length windows in = 2 samples and can be easily implemented digitally. The resolution in the frequency domain must be better than Δf = 16 Hz. This corresponds approximately to the distance of the transmission frequencies. For the required window length in:

- <16Hz m

The blurring between temporal and spectral resolution requires a tradeoff given by the equation

is described. The choice of in = 256 assuming fs = 1600 Hz leads to an observation window of Δt = 160 ms and a frequency resolution of Δf = 6.25 Hz. A doubling of in leads to an impermissibly long window and halving worsens the frequency resolution, so that a clean signal detection is difficult.

There are a variety of algorithms for calculating the FFT on a digital system. Many providers of DSPs or FPGAs offer ready-made IP cores, which are specially tailored to the target hardware. As a result, transformation times of a few μs are already possible on low-cost components. For such applications, however, such services are superfluous and it remains to be determined whether space can be saved at the expense of performance through an own implementation. The use of self-developed systems simplifies the case for admission and can help to reduce costs.

The output of the function block Transformation provides, independent of the selected transformation, a set of coefficients representing the original signal. In contrast to the other methods, the amplitude in the time domain is not averaged and evaluated, but the calculated coefficients are compared with threshold values. In the case of a signal transformed with the STFT, only three coefficients need to be examined in each interval. When the FFT has been calculated with the parameters given above, the coefficients 34, 37 and 40 represent exactly the proportions of the transmission frequencies. More information is usually not necessary for the transition decision. As explained above, in order to meet the time constraint for occupancy, the coefficients must be evaluated permanently. The detection of an isolator surge requires this continuous information as well because the adjacent track circuits are not synchronized. This ensures that the contact time of the neighboring circuit on the unused channel is sufficiently long. For the detection of all other operating cases, it suffices to evaluate the coefficients at certain points in time. The time te are chosen so that the FFT window is completely within a transmitter interval and consequently only one coefficient contains the information about the transmission signal.

A further advantage is that the temporal transition region occurring during run-time effects when switching between the channels can be avoided. For selective evaluation, the interval is used, which begins 10 ms after the switching of the transmission frequency. Figures 17, 19 and 20 illustrate the timing of the evaluation in the undisturbed state. The marked coefficients represent the values at the fixed times.

The FFT coefficients serve to trigger one of the transitions shown in FIG. They are calculated on an ongoing basis, but in most cases are only taken into account at specific points for the evaluation. The exceptions are the transition Tl and T3 in the Isolierstoss bridging. The respective active frequency is hereinafter referred to as channel A with the associated coefficient Ca. The frequency not currently being transmitted is channel B with the coefficient Cb. The frequencies of channel A and B are thus interchanged with each interval. The coefficients correspond to the reception level in the remaining methods and are also compared with threshold values.

The transition Tl becomes active when the GSK is in state S1 and the coefficients of the two set frequencies are valid:

Ca + Cb <P2

Due to the continuous evaluation, the time condition is maintained at all times. The occupancy should not be reported immediately after the detection, but should be monitored for the remaining 100 ms to reduce the influence of short-term disturbances. The request to the transition is not very restrictive because S2 is the safe state. In the event of a malfunction, it may become a short-term zen occupancy come. In the next fixed interval, however, the fault is detected reliably. This procedure is allowed. The transition T2 becomes active when the GSK is in state S2 and applies in two consecutive fixed intervals:

Ca> Pl and Cb <Ptr

T2 is therefore more restrictive than T5. It is ensured that at neither of the two frequencies used a disturbance can lead to an incorrect FM without it being detected.

Transition T3 becomes active when the GSK is in state S1 and one of the following occurs: Override is detected when

Ca> Pmax

in two consecutive intervals. It can be assumed that the level is too high on both frequencies. If the level is too high on only one frequency, it is a constructive interference with an interference voltage. The section can remain free, as occupancy detection via the undisturbed frequency is still possible.

A ballast disturbance is recognized when the coefficient of channel A is at two consecutive intervals:

P2 <Ca <Pl

To detect the Isolierstossüberbrückung the coefficient Cx of the unused frequency is evaluated. For detection, a long observation time is available. This should be exploited to ensure that the isolator surge is not due to a shorter disturbance from traction currents. The fault is detected when the coefficient of the unused frequency is often met. C _x> Piso

Since the neighboring GSKs are not synchronized, it must be ensured that the level is sufficiently high even in the event of an unfavorable shift between the intervals of the two GSKs. This is ensured by the continuous evaluation of the unused channel.

A Traktionsstromstörung manifests itself by:

Cb> Ptr

The Transition T4 describes the release of the GSK from the disorder into a regular occupancy. For this transition more than 250 ms of the transition Tl are available. To trigger T4, must be in two consecutive intervals

ca <P2 and cb <P2

be. This condition is necessary so that the system can not change permanently with each interval between S2 and S3. The additional observation time can be claimed, since when driving free is driven slowly and not at line speed.

T5: The transition T5 is triggered when the GSK is in state S2 and the same conditions as at T3 are fulfilled.

In the previous section, new methods have been developed which allow a track vacancy message in the event of an undisturbed event. In addition to this basic function they should be as immune to dynamic disturbances as possible. To accurately evaluate this property and to be able to compare it with the existing system, various perturbations are defined in the following sections. The characteristic of the disturbance can be determined by experience from measured values or by theoretical considerations. During the life cycle of the UGSK of Siemens Schweiz AG, numerous measurements were carried out directly on the track infrastructure as well as in the primary circuit of locomotives.

Such punctual measurements can be used to analyze basic properties of the expected disturbances. A representative overview of all possible sturgeon scenarios by all traction vehicles, trains and special operating cases is not possible with reasonable effort. The measurements made so far considered only parts of the expected disturbances. The following characteristics of the disturbances can be derived from the available measurement results:

The traction return currents of converter vehicles contain a high harmonic content at low speeds. However, the amplitudes clearly fall below the limits. In measurements below 15 kV / 16.7 Hz, there are no significant harmonics above 200 Hz. With regard to the 25 kV / 50 Hz network there are currently no measurements available. The traction return currents of converter vehicles contain at high speeds a comparable harmonic content as at low speeds. The amplitude of the fundamental wave is correspondingly larger due to the converted active power. In this case, the shares above 200 Hz are also negligible.

The effects of Zugsammeischiene are so far little or not examined. Therefore, the assumption is allowed that in addition to the fundamental and their harmonics a significant noise occurs. In contrast to the harmonic emissions of locomotives the Zugsammeischiene is not subject to limits. Vehicles with DC machines without power electronics do not generate disturbing harmonic components. They have no influence on the track circuits above 200 Hz. The influences of exceptional operating conditions have also been insufficiently investigated. These include, for example, driving over a protection route with the main switch on or a short-circuiting of the catenary. In these cases, short, high current pulses in the rail are conceivable. All systems in railway technology are designed for a service life of several decades. Advances in power electronics in recent years have shown that estimating future interference on these systems is not trivial. The installed power on a locomotive will not increase significantly for mechanical reasons, but especially freight trains are heavier and led with more traction vehicles.

The length of passenger trains can not be excessively increased due to the limited infrastructure. The power consumption via the Zugsammeischiene and the associated interference will increase due to the growing comfort and service demands of customers with certainty. Another trend is flexible, fast trainsets with a distributed drive concept, the performance of which can significantly exceed that of a single traction unit. In Switzerland, lacking vehicles also lacks meaningful measurements. Although the disturbances in the real operating environment will increase in intensity in the future, they will not significantly change their basic characteristics.

The basis for the performance analysis of the new methods are the test signals described in this section. They are chosen so that all theoretically possible faults are covered.

The rail currents as the cause of the interference signals are described in their limits by the main operator and in-house. 4 Ampere rms must not be exceeded near the transmission frequencies.

sinewave

As a disturbance, a sinusoidal signal having a single frequency is added to the transmission signal. All parameters of the signal are variable. The frequency should be in the passband of the filter. The amplitude is chosen in the examples so that an extinction of the transmission signal is possible. The calculation of the interference voltage from the maximum permissible rail current for long sections shows that the interference voltage can assume the same amplitude as the transmission signal.

The probability that this signal occurs in real rail operation is very low. Nevertheless, it is used as a test signal, as it is a challenge especially for the methods with continuous transmit signal.

sough

The influences of the Zugsammeischiene be simulated by a mean value free noise. In the passband of the receive filter random, different high interference components occur. The SNR is deliberately kept very small in order to reproduce the notable achievements.

Traktionsström

A traction current with the fundamental frequency of 16.7 Hz and the first 15 harmonics simulate the effects of a moving train. The specified transmission frequencies are next to the harmonics, but a too wide receive filter can include them. The noise can be described in the passband like a sinusoidal signal of a frequency because the filters of Method I and III are sufficiently narrow.

pulse

This signal is used to investigate how long a disturbance must at least act to produce an incorrect FM or BM. The results can be used to improve the strategy for troubleshooting in detailed planning. It is conceivable, for example, that a too short, reported occupancy is subsequently treated as a disorder.

To illustrate the operation, simulations are created using MATLAB. Figure 17 shows the basic operation of this method. The upper graph shows the sampled receive signal during 400 samples. At the marked times, the switching between the frequencies fl and f2 takes place. The 256 samples within the marked FFT window serve as input to the transformation. They are shown in the following illustrations.

The lower graph shows the associated time profile of the three selected coefficients of the FFT. Since the transmission period on a channel is longer than the FFT window, there are selected times at which only one frequency of the transmitter occurs in the transformation result. These times are marked.

Figure 18 shows the coefficients at the excellent times over a longer time. In the undisturbed state, the coefficient reaches approximately the normal value, while cb disappears. If this situation persists for at least two intervals, the section is certainly considered free. In addition to the recognition of the free state with the aid of the coefficients at the fixed times, the system recognizes an occupancy by means of the continuously calculated coefficients. A sufficiently good occupancy produces the course of the coefficients shown in FIG. From the 960th sample the section is occupied. During the next 160 ms, the coefficients are smaller and then the occupancy is detected, since both cl and c2 have fallen below P2. This situation can be monitored for another 100 ms before the occupancy is reported. These operating cases show that the evaluation can fulfill the basic function. The oversteer and ballast disturbance are detected with the same information and other thresholds. The Isolierstossüberbrückung as remaining static interference requires separate consideration.

The neighboring UGSK must be set so that they always differ in a transmission frequency. For the example in Figure 20, the simulated GSK uses frequencies 11 and 12, and the neighbor uses frequencies 12 and 13. After 400 samples, there is a shock bypass and the receiver also receives the signal from the neighboring GSK. The ratio of the Amplitudes from the parameters of the infrastructure. Their change is not taken into account in this example. The phase and switching times of the frequency are also random because the timing sources of the GSK are not synchronized.

The alternating use of two frequencies with identical transmission periods ensures, however, that the non-common frequencies of the two GSK act during at least half an evaluation interval and here the coefficient c3 becomes sufficiently large. The example shows this borderline case. For Sample 816, the coefficients are evaluated and there is c3 »0. If this event occurs too frequently in the 20-second observation time, the isolator surge can be reported.

In summary, it can therefore be stated that an advantageous embodiment of the method according to the invention evaluates the coefficients of a fast Fourier transformation of the received signal. It offers the following advantages:

The effective noise bandwidth of 12.5 Hz is comparable to that of Method I, which is a factor of 2 better than the existing system. For occupancy, no unfavorable times must be taken into account, which lead to compromises in immunity. Deletion of the transmission signal by a narrow-band interference is not possible. The changing transmission frequency creates artificial transmission pauses. They allow a comparable fault detection as the well-known in the art track circuit UGSK 95 Siemens Siemens AG.

The FFT is well studied and can be implemented digitally. The safety case can be provided closed. On the other hand, only the following minor disadvantages of this method speak: The FFT requires slightly more computing power than the three method I-FIR filters. The time limits for fault detection can only be kept to a minimum in the worst case scenario. However, these limits are not absolute limits, but are rather indicative. Operating parameters of the method are:

Parameter value

Oversampling frequency = 12.8 kHz

Sampling frequency Is = 1600 Hz

Transmitting frequency 1 fl = 208.1 Hz

Transmission frequency 2 f2 = 225.0 Hz

Transmitting frequency 3 f3 = 243.75Hz

Transmitter interval ta, b = 170 ms

FFT window length In = 256

FFT accuracy 16 bits

Table 5: Process parameters

Claims

claims

A method of track release of a portion of a track track, comprising: a) defining a length of the section via a bus bar interrupted at two ends; b) at one end of the rail line an alternating voltage with two alternating frequencies is fed as an input signal; c) an output signal is tapped off at the other end of the rail track; d) the tapped output signal is analyzed for its components in the two frequencies; and e) it is decided in dependence on the analysis by means of limit comparisons which state the track section has.

2. The method of claim 1, wherein the section is checked for the states "FREE", "BUSY" and "FAILED".

3. The method of claim 1 or 2, wherein the output signal is analyzed three channels.

4. The method according to any one of claims 1 to 3, wherein the output signal is bandpass filtered at least in two channels, wherein the pass frequencies of the band passes substantially equal to the two frequencies of the input signal.

5. The method of claim 4, wherein in a first channel, the level of the output signal is analyzed on the frequency of the input signal being transmitted, in a second channel, the level of the output signal is analyzed on the frequency of the currently unsent input signal and in a third channel the frequency spectrum of the output signal is analyzed.

6. The method according to any one of the preceding claims, in which for track release of two adjacent sections of a track section in the two sections, only one frequency of the input and the other frequency differs from each other, wherein the frequency currently being transmitted in the adjacent sections is also set different from each other.

A track-release system of a section of a rail track, comprising: a) a bus bar interrupted at two ends, the length of which defines the section length; b) a transmitter which at one end of the rail line an alternating voltage with two alternating frequencies fed as input to the rail track; c) a receiver with which an output signal is tapped at the other end of the rail line; d) an analysis device with which the tapped output signal is analyzed with respect to its components in the two frequencies; and e) a logic device which, depending on the analysis based on limit value comparisons, decides which state the track section has.

8. The system of claim 7, wherein the section is testable to the states 'FREE', 'BUSY' and 'INTERFERENCE'.

9. System according to claim 7 or 8, wherein the output signal is three-channel analyzable.

A system according to any one of claims 7 to 9, wherein the output signal is bandpass filtered at least in two channels, the pass frequencies of the bandpasses substantially corresponding to the two frequencies of the input signal.

11. The system of claim 10, wherein in a first channel, the level of the output signal on the frequency of the currently transmitted input signal is analyzable, in a second channel, the level of the output signal on the frequency of the currently unsent Input signal is analyzable and in a third channel, the frequency spectrum of the output signal can be analyzed.

12. System according to any one of the preceding claims 7 to 11, in which for the track release of two adjacent sections of a track section in the two sections only one frequency of the input signal matches and the other frequency differs from each other, wherein the respectively just transmitted frequency in the adjacent sections also different from each other is adjustable.