AU2021245132B2 - Method for identifying characteristics of a rail vehicle and an apparatus suitable for implementing the method - Google Patents

Abstract

Method for identifying characteristics of a rail vehicle and an apparatus suitable for implementing the method The subject matter of the invention is a method for identifying characteristics of a rail vehicle (FZ) in which an axle counter (AZ1... AZ2) captures measurement data while the rail vehicle (FZ) is passing over. The measurement data is analyzed in a computer-aided manner and in this case wheelbases of the rail vehicle (FZ) are determined. A first characteristic is determined in a computer-aided manner with reference to the determined wheelbases, namely whether the rail vehicle (FZ) is a passenger train or a freight train. In the case of determining the first characteristic in a first testing step, a check is performed as to whether at least in a predominant part of the sequence of the wheelbases it is possible to determine identical or similar patterns. If it has not been possible to determine a pattern, the characteristic of a freight train is allocated to the rail vehicle (FZ) as the first characteristic. If a pattern has been determined, the characteristic of a passenger train is allocated to the rail vehicle (FZ) as the first characteristic and/or a further testing step is performed so as to safeguard or to expand the testing result. This process can be supported by machine learning. Moreover, the invention comprises an apparatus for determining characteristics of rail vehicles (FZ), a computer program product and also a providing apparatus. Fig. 1 1/3 A 2 >) FIG 1 LZ S4 S5 BU 3 FZ A3 G A1 >))AZ1 cAZ2 GLS3-/ A1 \S1 -\S2 S6 SW FIG 2 PZ LK PW PW PW TK FF E F F G A B A C A B A C A D A MT=ABAC FIG 3 GZ LK GW1 GW2 GW3 A B A C D E D F G F H

Description

1/3 A2 >) FIG 1 LZ

S4 S5 BU 3 FZ A3

G A1 >))AZ1 cAZ2 GLS3-/ A1 \S1 -\S2

S6 SW

FIG 2 PZ

LK PW PW PW TK FF E FF G A B A C A B A C A D A

MT=ABAC

FIG 3 GZ LK GW1 GW2 GW3

A B A C D E D F G F H

2020P20998 1

TITLE

Method for identifying characteristics of a rail vehicle and

an apparatus suitable for implementing the method

BACKGROUND

The classic safety technology in rail operations does not

identify many characteristics of trains, such as for example

the type of a train (for example freight train, regional

train, locomotive), but rather only knows logical

characteristics, for example whether a train detection section

is occupied. The operation control technology on the other

hand identifies this type and where applicable further

characteristics of the train; these characteristics however

cannot in general form the basis of safe decisions because the

characteristics themselves do not have the required level of

safety. Nevertheless, it is required for the train operation

in specific operating situations to determine the type of

train with the required level of safety. This is provided

nowadays by a mixture of technical and operational methods, in

part with considerable outlay.

Train types, in Switzerland referred to as train categories,

are categories of different trains. The classification of the

trains is provided with regard to the use of said trains,

according to their meaning for travel and on account of their

rail operation management. Each train is referred to by the

train type and a train number.

The references for the train types vary; in addition to the

colloquial references there are therefore also technical

designations and namely transportation scientific terms,

references that have arisen via regulations from the time of

2020P20998 2

national railways and also brand names of the rail travel

companies. Regardless of which train types are used, these

train types however render it possible to provide more precise

information with regard to the traveling train. These

references can be stored for example in a rail automation

system and can be drawn upon for control tasks.

The document EP 2 718 168 B1 relates to a method for operating

a rail safety system having at least one trackside facility

while taking into consideration a speed measurement variable

that is detected when the rail vehicle travels into the

switch-on section of the rail safety system. When the rail

vehicle travels into the switch-on section a check is

performed with reference to the speed measurement variable as

to whether a correction time is to be set for the relay of a

message from the one trackside facility to an allocated rail

safety arrangement according to the speed measurement

variable. Afterward, a correction time that is set is checked

for whether this correction time is to remain effective in

dependence upon at least one further journey time determining

influence variable of the rail vehicle.

SUMMARY

It is an object of the present invention to overcome or

ameliorate at least one of the disadvantages of the prior art,

or to provide a useful alternative.

In one aspect, the invention provides a method for identifying

characteristics of a rail vehicle in which

• an axle counter captures measurement data while the rail

vehicle is passing over,

2020P20998 3

• the measurement data is analyzed in a computer-aided

manner and in this case wheelbases of the rail vehicle

are determined,

• a first characteristic is determined in a computer-aided

manner with reference to the determined wheelbases,

namely whether the rail vehicle is a passenger train or a

freight train.

wherein in the case of determining the first characteristic in

a first testing step, a check is performed as to whether at

least in a predominant part of the sequence of the wheelbases

it is possible to determine identical or similar patterns and

• if it has not been possible to determine a pattern, the

characteristic of a freight train is allocated to the

rail vehicle as the first characteristic, or

• if a pattern has been determined, the characteristic of a

passenger train is allocated to the rail vehicle as the

first characteristic and/or a further testing step is

performed.

In another aspect the invention provides an apparatus for

determining characteristics of rail vehicles, said apparatus

comprising

• at least one axle counter for capturing measurement data

when rail vehicles pass over,

• a computer that is configured so as to analyze the

measurement data and in this case to determine wheelbases

of the rail vehicle, and also

• with reference to the determined wheelbases to determine

a first characteristic as to whether the rail vehicle is

a passenger train or a freight train.

wherein the computer is moreover configured for the purpose

of, during the determination of the first characteristic in a

first testing step, checking whether it is possible to

2020P20998 4

determine a repeating pattern (MT) at least in a predominant

part of the sequence of the wheelbases that are determined and

• if it has not been possible to determine a pattern, to

allocate the characteristic of a freight train to the

rail vehicle as the first characteristic, or

• if a pattern has been determined, to allocate the first

characteristic of a passenger train to the rail vehicle

(FZ) and/or to perform a further testing step.

In yet another aspect, the invention provides a computer

program product having program commands for implementing the

first aspect of the invention.

In a further aspect, the invention provides a computer program

product and also a providing apparatus for this computer

program product, wherein the computer program product is

equipped with program commands for implementing this method.

In another aspect, the invention resides in identifying the

type of the train with a sufficient reliability if possible

without additionally installing sensors so that the identified

train type can be used in the safety technology of the rail

operation. For this purpose, a method and an apparatus that is

suitable for implementing the method is to be disclosed.

Moreover, the object of the invention resides in disclosing a

computer program product and also a providing apparatus for

this computer program product with which it is possible to

implement the above-mentioned method.

This aspect is achieved in accordance with the invention with

the claim subject matter (method) that is disclosed in the

summary by virtue of the fact that in the case of determining

the first characteristic in a first testing step, a check is

performed as to whether at least in a predominant part of the

2020P20998 4a

sequence of the wheelbases it is possible to determine

identical or similar patterns and

• if it has not been possible to determine a pattern, the

characteristic of a freight train is allocated to the

rail vehicle as the first characteristic, or

• if a pattern has been determined, the characteristic of a

passenger train is allocated to the rail vehicle as the

first characteristic and/or a further testing step is

performed.

In accordance with the invention, in other words it is

provided to estimate the intervals of the axles of the train

from the raw data of the axle counter while the train is

passing over. In particular, passenger trains such as ICE

(intercity express) or regional trains are embodied from fixed

units that in general remain together and it is not possible

to decouple any carriages from said fixed units. For this

reason, there is a pattern that can be measured by coupling

these units multiple times one behind the other and that are

similar to one another. Passenger trains thereby so to speak

have a fixed "fingerprint" that is only changed by measurement

errors etc.

The advantage of the use of a pattern identification in the

train operation is due to the fact that parameters of the

train operation such as for example closing times of a level

crossing or track clearances can be adapted flexibly to the

vehicles that are allocated the first characteristic owing to

the pattern identification. As a consequence, for example an

improved route utilization is achieved. Another example is the

reliable identification of risk focus points for which safety

measures can be introduced.

2020P20998 4b

Conversely, in the case of freight trains there are other

patterns that are generally variable and thereby not similar

or identical patterns depending on which and how many units are coupled together. The data can therefore be represented as multidimensional vectors and their components estimate the intervals between the axles, in other words axle 1 to axle 2 to axle n-i to axle n (in the case of n axles of the train, in reality up to 250).

The terms identical and similar are to be understood in their meaning in the sense of the pattern identification. This means that a comparison of patterns can lead to the fact that these patterns are evaluated as identical or similar (or even not as identical or not as similar, in other words not related). This evaluation is preferably provided in a computer-aided manner.

Patterns are understood as identical if all the testing criteria in the case of a pattern comparison lead to the result that the testing criteria are met. Since the testing criteria amount to measurement values, it is possible in this case to determine a tolerance interval for the measurement and the testing criteria can lie within said tolerance interval in order to be understood as identical.

Patterns are understood as similar if an evaluation of the testing criteria provides that these patterns at least predominantly correspond to one another. In this case, it is to be noted that similarity is consequently also present if the patterns are identical.

In order to implement the pattern identification, it is necessary to determine when the question can be affirmed that the criteria at least predominantly correspond to one another. In general, in this case for the identification of the mentioned first characteristic (freight train or passenger train) the following relationship applies. The stricter the criteria for the identification of similarity, the greater the probability that the identified similar patterns actually always lead to the identification of passengers. However, the probability that passenger trains are not identified also increases. The less strict the criteria for the identification of similarity, the higher the probability that all passenger trains are identified. However, the probability that freight trains are accidentally identified as passenger trains also increases.

Regardless of the strictness of the criteria, the method in accordance with the invention functions in the technical sense. During operation, however it is to be determined where an optimum lies with regard to the strictness of the criteria in relation to a safe operation (see below regarding the determination of the criteria).

In this case, it is to be taken into consideration that the fact that no at least similar patterns have been identified leads to the evaluation of the first characteristic such that the characteristic is assumed as freight train. It is to be noted that with regard to the train operation freight trains are to be evaluated more critically. For example, a tunnel encounter between two passenger trains, the outer dimensions of which can be reliably determined, could be permitted, however the tunnel encounter of two freight trains or a freight train with a passenger train could not be permitted. Another example are level crossings. A freight train on account of its slower speed will require longer closing times of the level crossing than would be required for a passenger train.

It is possible to derive therefrom that the method must be set so that in the event of doubt, the characteristic should be assumed as a freight train. For the pattern identification method this means that the train operation can be performed with a high safety standard (for example SEL four) if the criteria are set as strict. In other words that under no circumstances can a freight train be identified as a passenger train. On the other hand, it is less critical if a passenger train is not identified with reference to its characteristic pattern since its management as a freight train during the train operation is in general less non-critical.

A predominant part of the sequence of the wheelbases is

provided if it is possible to determine the repeating pattern

for more than 50 % of the wheelbases. It can also preferably

be defined that the limit at which a predominant part is to be

assumed is more than 60 %, particularly preferably more than

%, 80 % or 90 %.

The term "computer aided" or "computer implemented" can be

understood in the context of the invention as an

implementation of the method in which at least one computer or

processor performs at least one method step of the method.

The expression "computer" covers all electronic devices having

data processing characteristics. Computers can be for example

personal computers, servers, handheld computers, mobile radio

devices and other communications devices that process computer

aided data, processors and other electronic devices for data

processing that can preferably also be connected together to a

network.

The term "processor" can be understood in the context of the

invention to be for example a converter, a sensor for

generating measurement signals or an electronic circuit. A

processor can in particular be a main processor (in other

words a central processing unit, CPU), a microprocessor, a microcontroller, or a digital signal processor possibly in combination with a storage unit for storing program commands etc.

The term "processor" can also be understood as a virtualized processor or a soft CPU.

The term "storage unit" can be understood in the context of the invention for example as a computer readable storage device in the form of a main memory (in other words random access memory, RAM) or data storage device (hard disc or data carrier).

As "interfaces" it is possible to realize in a hardware technical manner for example wired or as a radio connection and/or in a software technical manner for example as interaction between individual program modules or program parts of one or multiple computer programs.

A "cloud" is to be understood as an environment for "cloud computing". An IT infrastructure is intended that is made available via interfaces of a network such as the Internet. The cloud generally contains storage space, computing power or software as a service without the need for this to be installed on the local computer that is using the cloud. The services that are offered within the scope of cloud computing comprise the entire spectrum of information technology and contain inter alia infrastructure, platforms and software.

"Program modules" are to be understood as individual functional units that render possible the program sequence in accordance with the invention. These functional units can be realized in an individual computer program or in multiple computer programs that communicate with one another. The interfaces that are realized in this case can be implemented in a software technical manner within a single processor or in a hardware technical manner if multiple processors are used.

Provided that in the following description nothing different is disclosed, the terms "create", "determine", "calculate", "generate", "configure", "modify" and the like preferably relate to processes that generate data and/or amend data and/or transfer the data into other data. In this case, the data is provided in particular as physical variables, for example as electrical pulses or also as measurement values. The required program command instructions are collected in a computer program as software. Furthermore, the terms "transmit", "receive", "input", "output", "transfer" and the like relate to the interaction of individual hardware components and/or software components via interfaces.

In accordance with one embodiment of the invention, it is provided that in the case of the first testing step in the sequence of the wheelbases a number of wheelbases remain as not taken into consideration at the start of the sequence and/or a number of wheelbases remain as not taken into consideration at the end of the sequence.

By not taking into consideration a number of wheelbases at the start or at the end of the sequence of wheelbases it is advantageously possible to realize that locomotives or end cars that for example in the case of passenger trains form the start or the end of the train, are not tested for the presence of patterns in the sequence of wheelbases. Both the locomotives as well as frequently also the end cars namely have different wheelbases (that are consequently also characterized by different patterns) than the vehicles in the middle of the train that are generally identical in the case of a passenger train and therefore form similar or identical patterns. The first testing step can therefore be performed in an advantageously more rapid and reliable manner if it is limited in particular to the middle part of the train.

The number of axles that are not to be taken into

consideration depends upon the train operation that is to be

monitored. If the train machines that are to be used are

known, the sequence corresponds to the wheelbases that are not

to be taken into consideration of any locomotives or end cars

that are used. However, it is also possible in the case of

unknown locomotives and end cars to assume a blanket value.

This can be for example four, six or eight axles.

In accordance with one embodiment of the invention, it is

provided that in the or in a further testing step the quantity

of wheelbases is determined, wherein the characteristic of a

passenger train is only then allocated to the rail vehicle as

the first characteristic if the quantity of the largest

wheelbase that is provided in the pattern exceeds a determined

limit value.

Which limit value reliably allows a conclusion of passenger

trains depends not least also on the features of the

respective train operation that is to be monitored. This limit

value can consequently be determined in dependence upon the

track if it is known which passenger trains travel on the

relevant track. In this case, it is important that the in each

case longest occurring wheelbase of the relevant train

carriage is taken into consideration.

However, for the case that train carriages having different

wheelbase lengths are used, the shortest of the in each case

longest intervals of the different train carriages of passenger trains that are traveling on the track is taken into consideration as the limit value.

This limit value in accordance with a particularly

advantageous alternative of the invention can further have a

difference with respect to typical wheelbases of freight

trains so that the wheelbase can be drawn upon as a

particularly reliable criterion for differentiating freight

trains. Since the wheelbase is used as an additional (in other

words supplementary) criterion to the patterns that are to be

identified, the adherence to this difference is however not

obligatory.

In accordance with one embodiment of the invention, it is

provided that in the or in a further testing step, the

determined patterns of the wheelbases are compared with

reference patterns of wheelbases and in the event of an

identified match of the pattern with a reference pattern a

train type that is linked to the reference pattern is

allocated to the rail vehicle as a second characteristic.

The reference pattern can be filed for example in a storage

device. A server can provide the reference pattern so that a

comparison with the determined patterns is rendered possible.

Another possibility resides in the fact that the reference

patterns are filed in a storage device that forms part of the

axle counter. The possibility is therefore provided of

technically modifying the axle counter with a specific

intelligence, in other words as units that operate in an

autarkic or in part autarkic manner.

The advantage in the fact that reference patterns are filed in

a storage device resides in the fact that these reference

patterns are available at any time and can be retrieved when required without a time delay. The storage devices can also store the different reference patterns in a track-specific manner so that only specific reference patterns are made available to specific axle counters at specific track sections.

In accordance with one embodiment of the invention, it is

provided that further characteristics of the rail vehicle, in

particular the direction of travel of the train and/or the

speed of the train when an axle is passing over and/or the

average speed when an axle is passing over and/or the

acceleration when an axle is passing over and/or the wheel

diameter are determined using the axle counter.

In addition to the detection of the number of axles, owing to

the conventional embodiment as a double sensor system the axle

counter is also fundamentally suitable for determining further

data such as those mentioned above. Moreover, it is possible

in a relatively simple manner to supplement the axle counter

with simple sensors, for example the axle load when an axle is

passing over.

It is possible to use the following measuring principles in an

exemplary manner.

• direction of travel of the train: by comparing the

influence in the case of double sensors (for example by

evaluating the time offset in the case of the signal

generation)

• speed of the train when an axle is passing over: from the

spacing of the double sensors for example by evaluating

the time offset in the case of the signal generation or

the time interval of the passage of the estimated wheel

center point in the case of known wheelbases.

• average speed when an axle is passing over and/or the acceleration when an axle is passing over: from the average of various wheels or numerical derivation of the speed • wheel diameter: from the temporal duration of an influence of the axle counter

A suitable set of parameters for the application case are advantageously selected from a set of parameters that are determined in this manner. For example, it is possible to omit the wheel diameter if it is more or less identical for all trains on the track. For the parameters in question, location specific, representative data is now collected or measured and classified, for example passenger train, freight train. This is a final number of integer or real significant measurement values, for example this could be the speed and the number of axles, in order to provide a clear two-dimensional example. In other words, in principle a classification object is obtained as described below with regard to Figure 5.

Altogether, the collection of further parameters in addition to the patterns that are to be compared makes the identification of characteristics of vehicles more robust with respect to errors. Advantageously, it is possible in the case of identifying the train to achieve a higher degree of reliability so that it is possible to control train traffic in a more effective manner. Which parameters are to be taken into consideration in the case of the current control task for train travel then depends on the circumstances of the individual case. The parameters are to be selected in a suitable manner when the control method is conceived.

In accordance with one embodiment of the invention, it is provided that the criteria for the first testing step and/or the further testing steps are evaluated using a method of machine learning.

The machine learning advantageously renders it possible to

optimize the processes that are running, in other words the

reliable identification of the train characteristics, in

particular train types, during the operation. As a

consequence, the system can also automatically adapt to

changing operating conditions. For example, it is possible to

create additional patterns if a new type of passenger train is

used on a specific track section. For example, neural networks

or also other facilities having artificial intelligence can be

used for this purpose.

The term "artificial intelligence" (also referred to below as

AI) is to be understood within the scope of this invention in

the narrower sense to be the computer aided machine learning

(machine learning is also abbreviated to ML below). In this

case, this is the statistical learning of the parameterization

of algorithms, preferably for complex application cases. By

means of ML, the system identifies and learns patterns and

regularity in the detected process data with reference to

previously input learning data. With the aid of suitable

algorithms, it is possible by ML to find dedicated solutions

to problems that arise. ML divides itself into three fields

supervised learning, unsupervised learning and reinforcement

learning, with specific applications, for example regression

and classification, structural identification and prediction,

data generation (sampling) or autonomous tasks.

During supervised learning, the system is trained by the

relationship of input and associated output of known data and

learns in this manner approximative functional relationships.

In this case, the availability of suitable and sufficient data is crucial since if the system is trained using unsuitable (for example non-representative) data, the system thus learns erroneous functional relationships. During unsupervised learning, the system is likewise trained using example data, however only using input data and without the relationship to a known output. The system learns how to form data groups and to expand what is typical for the application case and where deviations or anomalies occur. As a consequence, the application cases can be described and error states can be discovered. During reinforced learning, the system learns via trial and error in that it proposes solutions to given problems and obtains a positive or negative evaluation with regard to this proposal via a feedback function. Depending upon the reward mechanism, the AI system learns to perform corresponding functions.

The machine learning can be performed for example by artificial neural networks (referred to in short below as ANN for artificial neural network). Artificial neural networks are usually based on the networking of many neurons, for example McCulloch-Pitts-Neurons or minor deviations therefrom. Fundamentally, other artificial neurons can also be used in ANN, for example the high order neuron. The topology of a network (the allocation of connections to nodes) must be determined in dependence upon the task of said network. The training phase in which the network "learns" follows after the construction of a network. In this case, a network can learn for example by the following methods:

• development of new connections

• deleting existing connections

• changing the weighting (the weighting of neuron j to neuron i)

• adapting the threshold values of the neurons provided that these neurons have threshold values

• adding or deleting neurons

• modifying the activation function, propagation function or output function

Moreover, the learning behavior changes in the case of changing the activation function of the neurons or the learning rate of the network. Viewed practically, an ANN learns mainly by modification of the weighting of the neurons. An adaptation of the threshold value in this case can be provided by an on neuron. As a consequence, ANN are capable of learning complicated non-linear functions via a learning algorithm that tries to determine all the parameters of the function by an iterative or recursive approach from input values that are provided and desired output values. ANN are in this case a realization of the connectionist paradigm since the function is embodied from many simple identical parts. Only in their sum does the behavior become complex.

In accordance with one embodiment of the invention, it is provided that probability densities for the characteristics are determined from the measurement data of a plurality of measurements.

The knowledge of the probability densities renders it possible to define classification limits for the allocation of the characteristics. In this case, the method is advantageously particularly robust with regard to the classification limits since in the case of the comparatively low dimensional problems in accordance with the invention it is possible to estimate from the data the probability densities for the two classes (for example using density estimation of the measurement results) and thereby also to determine the error probabilities for an incorrect classification.

In accordance with one embodiment of the invention, it is

provided that the method is implemented in order to determine

the characteristics of the rail vehicle while this rail

vehicle is approaching a risk site on the track that is

provided for said rail vehicle.

A risk site could be a site on the track in which either the

rail vehicle is potentially at risk (for example a tight

curve, bridge, tunnel) or also a site in which the rail

vehicle potentially risks others (for example a track

construction site or a level crossing).

The method in accordance with the invention can consequently

advantageously also be implemented selectively in order to

diffuse risk sites. In other words, train traffic can pass

risk sites in a more reliable manner while simultaneously

ensuring a higher degree of flexibility. In particular, the

trains can be guided through a risk site at different speeds

depending upon the characteristics of said trains. In other

words, that passenger trains for example can pass the risk

site at a higher speed than freight trains. This renders it

possible to more smoothly process the passenger train traffic.

The passengers in this manner arrive earlier at their

destination.

In accordance with one embodiment of the invention, it is

provided that the method is implemented for rail vehicles that

are approaching a level crossing, wherein the closing time of

the level crossing is selected in dependence upon the

determined rail vehicle if characteristics of the approaching

rail vehicle have been determined.

The great advantage in the case of using the method in the case of level crossings as risk sites lies in the fact that closing times of the level crossing can be advantageously set individually in dependence upon the characteristics of the approaching train. At least if it has been possible to reliably determine the characteristics of the train, in many cases the closing time can be shortened without this leading to a reduction in safety standards during the operation of the level crossing. There are benefits for the crossing traffic that in many cases has to wait for less time at the level crossing.

In accordance with one embodiment of the invention, it is provided that for a determination of the closing time a data pool is used in which closing times are linked to the determinable characteristics of the rail vehicles, in particular train types.

The data pool can be determined in a deterministic manner and/or can be created and/or further developed with the aid of the already explained above methods of machine learning during the operation. As soon as the data is provided in the data pool, it is possible to use said data advantageously with short access times. During the operation, the data in the data pool can be further optimized so that the train operation is increasingly rationalized.

In accordance with one embodiment of the invention, it is provided that for the case that it was not possible to determine the characteristic of the rail vehicle, a standard closing time is selected for the level crossing.

In accordance with the invention any closing time that can reliably prevent a risk to the crossing vehicle and pedestrian traffic independently of the characteristics of the trains that are traveling on the track is to be understood as the standard closing time for the level crossing. The slow traveling freight trains that require the longest time from the trigger point of the track safety system to the level crossing and therefore require the longest closing time are critical for this purpose. This can consequently be defined as the standard closing time.

The advantage in the mentioned use of the standard closing time resides in the fact that on the one hand a without exception safer operation of the level crossing can be ensured and on the other hand a flexible adaptation of the closing times can be provided if it has been possible to determine the characteristics of the approaching train with sufficient reliability.

The mentioned object is alternatively also achieved in accordance with the invention with the claim subject matter (apparatus) that is disclosed in the introduction by virtue of the fact that the computer is configured for the purpose of, during the determination of the first characteristic in a first testing step, checking whether it is possible to determine a repeating pattern at least in a predominant part of the sequence of the wheelbases that are determined and • if it has not been possible to determine a pattern, to allocate the characteristic of a freight train to the rail vehicle as the first characteristic, or • if a pattern has been determined, to allocate the first characteristic of a passenger train to the rail vehicle and/or to perform a further testing step.

With the apparatus it is possible to achieve the advantages

that have already been explained in relation to the method

that is further described above. The statements made with

regard to the method in accordance with the invention also

apply accordingly for the apparatus in accordance with the

invention.

Furthermore, a computer program product is claimed having

program commands for implementing the mentioned method in

accordance with the invention and/or the exemplary embodiments

of said method, wherein in each case the method in accordance

with the invention and/or the exemplary embodiments of said

method can be implemented by means of the computer program

product.

Furthermore, a providing apparatus is claimed for storing

and/or providing the computer program product. The providing

apparatus is for example a storage unit that stores and/or

provides the computer program product. Alternatively and/or in

addition, the providing apparatus is for example a network

service, a computer system, a server system, in particular a

distributed, for example cloud based, computer system and/or a

virtual computer system that stores and/or provides the

computer program product preferably in the form of a data

flow.

The provision is performed in the form of a program data block

as a file, in particular as a download file, or as a data

flow, in particular as a download data flow, of the computer

program product. This provision can also be provided for

example as a partial download that is embodied from multiple

parts. Such a computer program product is input into a system

for example while using the providing apparatus so that the

2020P20998 21

method in accordance with the invention is loaded onto a

computer for implementation.

Further details of the invention are described below with

reference to the drawing. Identical or corresponding drawing

elements are in each case provided with identical reference

numerals and are only explained multiple times insofar as

differences occur between the individual figures.

The following explained exemplary embodiments are preferred

embodiments of the invention. In the case of the exemplary

embodiments, the described components of the embodiments

represent in each case individual features of the invention

that are to be considered independently of one another and

said features in each case develop the invention independently

from one another and thereby also are to be regarded

individually or in another combination than the illustrated

combination as a component of the invention. Furthermore, the

described components can also be combined with the above

described features of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows an exemplary embodiment of the apparatus in

accordance with the invention having the cause-effect

relationships of said apparatus,

Figure 2 and 3 show schematically part of an identical or

similar pattern of wheelbases for a passenger train and a

freight train,

Figure 4 shows an exemplary embodiment of the method in

accordance with the invention as a flow diagram, wherein the

2020P20998 22

function units and interfaces in accordance with Figure 1 are

indicated in an exemplary manner,

Figure 5 shows symbolically for two normal distributions for

determined measuring data, however this functions in principle

for all distributions.

PREFFERED EMBODIMENT OF THE INVENTION

A track system having a track GL, a control center LZ and a

signal tower SW are illustrated in Figure 1. A vehicle FZ in

the form of a train is driving on the track GL toward a level

crossing BU. A first axle counter AZ1 and a second axle

counter AZ2 are installed on the track GL and said axle

counters are configured so as in a known manner per se to

count the axles of the vehicle FZ.

The axle counter AZ1 is connected via a first interface S1 and

also the second axle counter AZ2 is connected via a second

interface S2 to the signal tower SW, strictly speaking to a

computer CP that is provided in this signal tower. Moreover,

the computer CP has a third interface S3 for the level

crossing BU. Moreover, the computer CP is connected via a

sixth interface S6 to a storage unit SE.

The signal tower SW has a first antenna system Al, the control

center LZ has a second antenna system A2, and the vehicle FZ

has a third antenna system A3. As a consequence, both the

communication of the signal tower SW via a fourth interface S4

with the control center LZ as well as the communication of the

vehicle FZ via a fifth interface S5 with the control center LZ

are possible. The fourth interface S4 and the fifth interface

S5 are in this respect radio interfaces. The first interface

Sl, the second interface S2 and also the third interface S3

2020P20998 22a

can represent both wired as well as radio interfaces, wherein

for the latter case the antenna technology that would be required for the embodiment of radio interfaces is not illustrated.

If the vehicle FZ moves on the track GL toward the level

crossing BU, the axles of the vehicle FZ initially pass the

second axle counter AZ2 and subsequently the first axle

counter AZ1. The measured values that are recorded can be

transmitted via the first interface Si and the second

interface S2 to the computer CP, wherein the computer CP is

configured so as to implement the method in accordance with

the invention. In this case, the computer CP can also directly

assume control of the level crossing BU. Another possibility

resides in the fact that the computer CP is connected via the

third interface S3 to a further computer (not illustrated in

Figure 1) that is used to control the level crossing BU via a

further interface.

Figure 2 illustrates a passenger train PZ that is traveling on

the track GL as a vehicle FZ in accordance with Figure 1. This

passenger train PZ is embodied from a locomotive LK, multiple

passenger carriages PW and an end car TK on the end of the

passenger train PZ that lies opposite the locomotive LK.

Furthermore, the wheelbases between the individual axles

(indicated by wheels) are illustrated schematically. It is

shown that various wheelbases are provided repeatedly in the

passenger train PZ so that the sequence of the wheelbases can

be assessed for the presence of patterns. The wheelbases are

identified using upper case letters A to G. The sequence of

the wheelbases is FFEFFGABACABACABACADA.

If the locomotive LK and the end car TK are set aside since

these differ from the passenger carriages PW with regard to

their wheelbases, a sequence of wheelbases is therefore provided for the passenger carriages that are identical in construction and that follow one another and said sequence of wheelbases is always repeated. These wheelbases in this respect form a pattern MT for the passenger carriage PW that is following the locomotive LK and said pattern is identified using a curved bracket. The sequence of the wheelbases in the pattern MT that is illustrated in Figure 2 is ABAC. This sequence of wheelbases is also provided for the two subsequent passenger carriages.

A different situation is provided in the freight train GZ that

is illustrated in Figure 3 on the track GL and said freight

train GZ is embodied from a locomotive LK and also a first

freight carriage GW1, a second freight carriage GW2 and a

third freight carriage GW3. These carriages have different

lengths and number of axles so that multiple different

wheelbases are provided that are provided with the upper case

letters A to H. In Figure 3 it becomes clear that in the

illustrated sequence ABACDEDFGFH it is not possible to

discover any repeating pattern, which renders it possible to

conclude a freight train.

Figure 4 illustrates how the method in accordance with the

invention can be implemented. Initially, this method is

started in a first step START. A measuring step MS by the

relevant axle counter AZ1, AZ2 (cf. Figure 1) follows. A first

testing step PS1 follows this measuring step and the sequence

of the wheelbases in said testing step (as described with

regard to Figure 2 and Figure 3) can be determined and

checked. In this case, it is either possible to identify a

pattern MT in the sequence of the wheelbases or not. In a

following querying step GZ,PZ? a check is performed as to

whether owing to the sequence of wheelbases (by finding

patterns) it is possible to conclude a freight train GZ or a passenger train. If this is not the case, a standard closing time SZS for the level crossing BU is output to the storage device SE. A separate storage area is reserved in the storage device for this purpose and a controller (for example the computer CP or a further computer that is not further illustrated in Figures 1 to 3) of the level crossing can access said storage area in order to retrieve the prevailing stored closing time.

If it has been possible to determine the first characteristic,

in other words the question of whether it is a freight train

GZ or a passenger train PZ, a further query step GZ? follows

in the computer CP as to whether it is a freight train GZ. If

this is the case, a first calculated closing time SZ1 is

passed to the storage unit SE (replacing a previously stored

closing time). If this is not a freight train or a clear

result is not provided, a second testing step PS2 is

implemented in the computer CP.

The second testing step PS2 is used so as to determine the

quantities of wheelbases. In a following testing step |i<GW it

can therefore be queried as to whether the determined

quantities of the wheelbases are lower than a limit value that

is typical for freight trains GW. If this is the case, it is a

freight train GZ so that the first closing time SZ can be

passed to the storage unit SE (replacing a previously stored

closing time). If this is not the case, a third testing step

PS3 is initiated in the computer CP.

In the third testing step PS3, reference patterns RMT are

loaded from the storage unit SE. The wheelbases or the

quantities of said wheelbases are compared with the reference

patterns, wherein in a testing step MT=RMT it is possible to

test whether the determined patterns MT correspond to a reference pattern RMT. If this is not the case, a second closing time SZ2 is passed to the storage unit SE (replacing a previously stored closing time) and said second closing time can be understood as a standardized closing time for passenger trains PZ. However, if a pattern MT has been identified, it is therefore possible to pass a third closing time SZ3 to the storage unit SE (replacing previously stored closing time) and said third closing time is individually suitable for the reference pattern RMT. This individual third closing time SZ3 can already have been stored in the storage unit for example with the reference patterns RMT so that the handover of said third closing time to the above-mentioned separate storage area of the storage unit SE is performed on the basis of the data that is already available in the storage unit SE.

If the querying step MT=RMT turns out to be negative, the

determined pattern MT can also be passed to the control center

LZ via the interface S4. Simultaneously, travel data FD can

also be passed on from the vehicle FZ via the fifth interface

S5 to the control center LZ. It is possible with reference to

the data that is provided in the control center LZ to then

determine in a modification step MOD a new fourth closing time

SZ4 that is adapted to the determined train type and is passed

to the storage unit SE via an output step OUT. This fourth

closing time SZ4 can then be used as an individual closing

time for the level crossing BU (replacing a previously stored

closing time). Simultaneously, it is possible to output into

the storage unit SE such that the fourth closing time SZ4

together with the newly determined reference pattern RMT that

is associated with the vehicle FZ that is currently analyzed

is written in the storage unit SE as an addition into the data

bank.

A closing time for the level crossing BU is now in the storage

unit SE in the separate storage area. Depending on the

implementation of the method, this can be the standard closing

time SZS, the first closing time SZ1, the second closing time

SZ2, the third closing time SZ3 or the fourth closing time SZ4

(or further closing times that are not described in the

example in accordance with Figure 4).

This closing time is now available in the separate storage

area of the storage unit SE in order to be passed by

controlling the level crossing BU (cf. Figure 1), in other

words to the computer CP or to another controller of the level

crossing BU. The level crossing BU in this respect can be

operated using an individually determined closing time.

Figure 5 illustrates in an exemplary manner two parameters

that are illustrated in a plane that could also be referred to

as the x-y plane and said parameters in accordance with the

invention are measured or can be determined by the axle

counters and the measurement value distribution MV of the

measured values becomes clear on said plane. Accordingly the

speed GSW would be illustrated on the x-axis and the

wheelbases A... H would be illustrated on the y-axis. The z

axis is used so as to illustrate the (for example estimated)

probability densities.

For the parameters in question, in this example location

specific, representative data is obtained or measured and

classified, for example passenger train as normal distribution

NV2 and freight train as normal distribution NV1, as already

described above. This is a final number of integer or real

significant measurement data of the axle counter, for example

this could be speed and the wheelbase in order here to provide

a clear two-dimensional example. In other words, in principle a classification task is obtained, as is illustrated schematically in Figure 5.

If representative data is available, it is known how to solve

such problems of the pattern identification using methods of

machine learning, for example neural networks. In this case,

in this application in the case of axle counters a large

margin is provided for how the classification limits are set

since in the case of such low dimensional problems, it is

possible from the data to also estimate the probability

densities for the two classes (for example using density

estimation). It is thereby possible to determine the error

probabilities for an incorrect classification (cf. for example

Duda et al.: Pattern Classification, Wiley, 2001); Figure 5

illustrates this symbolically for a first normal distribution

NV1 and a second normal distribution NV2; however this

functions in principle also for other distributions as normal

distributions.

If in the example it is assumed that the small ellipse would

be the first classification limit KG1 for freight trains and

the large ellipse would be the classification limit KG2 for

passenger trains, then it could be possible using the

estimated distributions to calculate the error probabilities.

In the event of the error classification probability being too

high for freight trains, the classification limits would be

changed. In the example in accordance with Figure 5, a smaller

ellipse would have then been obtained for the first

classification limit KG1. There are however also applications

where the classification errors are asymmetrical, in other

words the errors do not have the same importance. For example,

in the case of a time controlled switch-on of a level

crossing, it would be irrelevant to take into consideration

the safety if a slower freight train would be classified as a quicker passenger train, while this would be dangerous in the event of a forbidden tunnel encounter. In other words, in each case the safety aspect must be taken into consideration in the case of the evaluation of the error types or error probabilities.

List of reference numerals

GL Track FZ Rail vehicle BU Level crossing LZ Control center SW Signal tower Al... A3 Antenna AZ1... AZ2 Axle counter S1... S6 Interface CP Computer SE Storage unit

PZ Passenger train LK Locomotive PW Passenger carriage TK End car GZ Freight train GWl... GW3 Freight carriages A... H Wheelbase MT Pattern RMT Reference pattern GW Limit value

MS Measuring step PS1... PS3 Testing step SZS Standard closing time SZ1... SZ4 Closing time (calculated) FD Travel data IN Input step MOD Modification step OUT Output step GZ? Querying step freight train?

MV Measurement value distribution GSW Speed KG1 . . KG2 Classification limit NV1 . . NV2 Normal distribution

Claims (14)

2020P20998 32 Claims

1. A method for identifying characteristics of a rail vehicle

in which

• an axle counter captures measurement data while the rail

vehicle is passing over,

• the measurement data is analyzed in a computer-aided

manner and in this case wheelbases of the rail vehicle

are determined,

• a first characteristic is determined in a computer-aided

manner with reference to the determined wheelbases,

namely whether the rail vehicle is a passenger train or a

freight train,

wherein in the case of determining the first characteristic in

a first testing step, a check is performed as to whether at

least in a predominant part of the sequence of the wheelbases

it is possible to determine identical or similar patterns and

• if it has not been possible to determine a pattern, the

characteristic of a freight train is allocated to the

rail vehicle as the first characteristic, or

• if a pattern has been determined, the characteristic of a

passenger train is allocated to the rail vehicle as the

first characteristic and/or a further testing step is

performed.

2. The method as claimed in claim 1, wherein in the case of

the first testing step in the sequence of the wheelbases a

number of wheelbases remain as not taken into consideration at

the start of the sequence and/or a number of wheelbases remain

as not taken into consideration at the end of the sequence.

3. The method as claimed in one of the preceding claims,

wherein in the or in a further testing step the quantity of

2020P20998 33

wheelbases is determined, wherein the characteristic of a

passenger train is allocated to the rail vehicle as the first

characteristic as long as the quantity of the largest

wheelbase that is provided in the pattern exceeds a specified

limit value.

4. The method as claimed in one of the preceding claims,

wherein in the or in a further testing step the patterns of

the wheelbases are compared with reference patterns of

wheelbases and in the event of an identified match of the

pattern with a reference pattern a train type that is linked

to the reference pattern is allocated to the rail vehicle as a

second characteristic.

5. The method as claimed in one of the preceding claims,

wherein further characteristics of the rail vehicle, in

particular the direction of travel of the train and/or the

speed of the train when an axle is passing over and/or the

average speed when an axle is passing over and/or the

acceleration when an axle is passing over are determined using

the axle counter.

6. The method as claimed in one of the preceding claims,

wherein the criteria for the first testing step and/or the

further testing steps are evaluated using a method of machine

learning.

7. The method as claimed in one of the preceding claims,

wherein

probability densities for the characteristics are determined

from the measurement data of a plurality of measurements.

8. The method as claimed in one of the preceding claims,

wherein the method is implemented in order to determine the

2020P20998 34

characteristics of the rail vehicle while this rail vehicle is

approaching a risk site on the track that is provided for said

rail vehicle.

9. The method as claimed in claim 8, wherein the method is

implemented for rail vehicles that are approaching a level

crossing, wherein the closing time of the level crossing is

selected in dependence upon the determined rail vehicle if

characteristics of the approaching rail vehicle have been

determined.

10. The method as claimed in claim 9, wherein for a

determination of the closing time a data pool is used in which

closing times are linked to the determinable characteristics

of the rail vehicles, in particular train types.

11. The method as claimed in one of claims 9 or 10, wherein

for the case that it was not possible to determine the

characteristic of the rail vehicle, a standard closing time is

selected for the level crossing.

12. An apparatus for determining characteristics of rail

vehicles, said apparatus comprising

• at least one axle counter for capturing measurement data

when rail vehicles pass over,

• a computer that is configured so as to analyze the

measurement data and in this case to determine wheelbases

of the rail vehicle, and also

with reference to the determined wheelbases to determine

a first characteristic of whether the rail vehicle is a

passenger train or a freight train,

wherein the computer is moreover configured for the purpose

of, during the determination of the first characteristic in a

first testing step, checking whether it is possible to

2020P20998 35

determine a repeating pattern at least in a predominant part

of the sequence of the wheelbases that are determined and

• if it has not been possible to determine a pattern, to

allocate the characteristic of a freight train to the

rail vehicle as the first characteristic, or

• if a pattern has been determined, to allocate the first

characteristic of a passenger train to the rail vehicle

and/or to perform a further testing step.

13. A computer program product having program commands for

implementing the method as claimed in any one of claims 1

11.

14. A providing apparatus for the computer program product as

claimed in claim 13, wherein the providing apparatus stores

and/or provides the computer program product.

Siemens Mobility GmbH

Patent Attorneys for the Applicant/Nominated Person

SPRUSON & FERGUSON