DESCRIPTION

The invention relates to a method for surveying the rail-wheel contact of a railway vehicle.

Wheel integrity is an essential part of the safe operation of railway vehicles. In particular in the case of high-speed trains, wheel damage has to be detected early and appropriately counteracted to avoid safety problems. On the other hand, overdiagnosis of wheel damage can lead to unnecessary wheel replacements which incur avoidable maintenance costs.

A generally known method in wheel maintenance is the so-called drive-by inspection, where the wheels are monitored while an inspection vehicle drives by. This is usually insufficient and leads to misdiagnoses, overlooking damaged wheels and marking serviceable wheels for replacement.

A more accurate method is the use of so-called wheel impact load detectors (WILDs). Such detectors consist of a series of strain gages welded to the rail quantifying the force applied to the rail on the basis of a mathematical relationship between the applied load and the deflection of the foot of the rail.

This impact forces thus measured can be applied for structural health monitoring of railway vehicle wheels. Currently, the impact load limit for wayside detectors is set at 90000 pounds by the Association of American Railroads.

Wheels exerting such forces, so-called high-impact wheels, often have flat spots on their tread surface, known as slid flat. Slid flats are usually caused when a wheel is blocked while the train is in motion, e.g. because of an engaged hand brake. Other damage modes, such as major defects in the tread surface, can also lead to high-impact wheels.

While high-impact wheels are primarily of concern because of the possibility of catastrophic failure with subsequent derailment, they are also of economic importance, since high-impact events place significant strain on the track. For example, high impact wheels have been observed to increase surface crack growth rate on rails by a factor of 100 compared to non-impact loading conditions and also have a detrimental effect on concrete sleeper health by increasing crack initiation.

Unfortunately, there is no easily established connection between wheel damage and failure. While some high-impact wheels can remain in service for years without failure, others fail almost immediately. Furthermore, certain modes of wheel failure, such as shattered rim failures and vertical split rim failures tend to occur on wheels well below the 90000 pound impact load limit.

It is therefore the objective of the present invention to provide a method for surveying the wheel-rail contact of a railway vehicle which allows for improved detection of wheel damage.

This objective is reached by a method according to claim 1 .

The method according to the invention comprises recording vertical and/or lat- eral acceleration of at least one wheel of the vehicle, storing the recorded acceleration together with an associated rotational position of the wheel, identifying acceleration events exceeding a predetermined parameter, and, for each identified event, classifying the event using a computational physical model of the wheel.

In other words, acceleration data, which is generally available from sensors inte- grated into axle bearings or the like, is employed for online monitoring of wheel health. Associating recorded acceleration data with the corresponding rotational positions of the wheel under surveillance allows for determining the exact position of acceleration- causing defects on the wheel's circumference and, furthermore, for determination and classification of defect types by means of a physical model of the wheel.

The methods enables railway operators to detect wheel defects as soon as they physically manifest themselves in the wheel's rolling behavior, thereby greatly increasing operational security and reducing maintenance cost due to a minimal rate of falsely positive detection events.

In a preferred embodiment of the invention, a frequency analysis of events asso- ciated with given rotational positions of the wheel is performed, and, if the frequency of at least one given event at one associated rotational position exceeds a predetermined threshold, the wheel is inspected. This helps to distinguish one-time or rare events caused e.g. by track defects or foreign object impact from actual wheel damage, thereby further reducing the probability of falsely classifying an operation wheel as faulty.

In a further preferred embodiment of the invention, the vertical and/or lateral acceleration of the wheel is stored together with an associated geographical position of the wheel. Geocoding acceleration events in that manner allows for simultaneous monitor- ing of wheel and track health, identifying sections of track in need of maintenance or inspection. For obtaining the geographical data, GPS may be employed.

It is further advantageous to calculate a diagonal acceleration from the vertical and lateral accelerations and store it together with the associated rotational and/or geo- graphical positions. This reduces the data set that needs to be handled and stored.

The acceleration and the associated rotational and/or geographical position can be stored locally in a train-mounted storage unit. Said data can then be read and analyzed during scheduled maintenance.

Alternatively, the acceleration and the associated rotational and/or geographical position can be transmitted to a data processing and storage unit external to the train. This is particularly advantageous if one is interested in on-line monitoring of track health, as the acceleration data for given geographical positions can be monitored over a whole fleet of trains in real-time, presenting an ever-accurate picture of track status.

In a further preferred embodiment of the invention, acceleration events are iden- tified by a pattern matching algorithm. This allows for accurate determination of a wide array of damage patterns, which can be classified exactly. Pattern matching furthermore allows for the detection of non-predetermined error types, which can subsequently be investigated.

In the following, the invention and its preferred embodiments will be explained in detail with reference to the drawings, which show:

FIG 1 a schematic representation of a railway vehicle wheel with associated sensors for acceleration and wheel position;

FIG 2 a flow scheme for an embodiment of the method according to the invention to determine wheel damage;

FIG 3 a flow scheme for an embodiment of the method according to the invention to additionally determine track damage and

FIG 4 a flow scheme for an embodiment of the method according to the invention for simultaneous determination of wheel and track damage.

In order to watch its status, a railway wheel 10 has attached sensors 12, 14 for measuring wheel acceleration and rotational position respectively.

The curve 16 shown in FIG 2 represents the course of wheel vibrations - as measured by the accelerometer 12 - over time. Together with curve 18, representing the rotational position of the wheel 10, these data form the basis for wheel damage determination. Pattern matching techniques are employed to detected impact events 20 within the vibration curve 16, which exceed certain predetermined parameters. In combination with a physical model 22 of the wheel, such events 20 can be classified, e.g. for failure type, and subsequently analyzed using frequency statistics, as shown in diagram 24. If the frequency of a certain type of event associated with a certain rotational angle of the wheel 10 exceeds a threshold, maintenance of the associated wheel is necessary.

As shown in FIG 3, such an analysis can be performed not only for the wheel 10, but also for the track 1 1 on which said wheel is running. To this end, impact events 20 are not only analyzed in dependency of wheel rotational position, but additionally in dependence of geographical position determined by a global positioning system client 26. The frequency analysis is then performed plotting event frequency against geographical position on the track, as visualized in diagram 28. If the impact event frequency exceeds a certain threshold for a particular segment of track, this segment is marked for inspection.

FIG 4 finally shows an integrated picture of the whole system, comprising a vehicle based data recording and monitoring device 30 which records and stores acceleration data from the wheels 10 and communicates these data to the physical model 22. Oh the detection of possible wheel damage, a warning signal 32 is generated and presented to the train operator.

A separate railway monitoring system 34 also receives the processed impact events from the physical model 32 and correlates them with geographical data received from the GPS receiver unit 26. Depending on current demands, the system 34 generates a list 36 of rail segments suspected of defects.

List of reference signs

10 wheel

1 1 track

12 accelerometer

14 rotational position sensor

16 curve

18 curve

20 event

22 model

24 frequency analysis

26 GPS receiver unit

28 frequency analysis

30 recording and storage unit

32 warning signal

34 railways monitoring system

36 list