The invention relates to a plant for the measuring of imbalance in wheels on vehicles running on rails, chiefly train wheels, which pass a measuring section on the rail element.

Uneven distribution of the mass on train wheels is an undesired characteristic which can have detrimen- tal effects on both rails and wheel bearings. There _^ fore, attempts are made to reduce such uneven mass distributions by regular machining of the wheels. This can result in unnecessary machining of the wheels and herewith extra operational costs and un- necessary reduction of the lifetime of the wheels. '

The uneven distribution of the mass can arise as a consequence partly of wear on the wheels, and partly as a consequence of great differences in temperature and high static loads on the wheels or combinations hereof. For some types of trains, wheel imbalance is especially highly troublesome. For example express trains where high rates of wheel rotation can give rise to harmful vibrations, specially-constructed, low train carriages with small wheels for the trans¬ port of fully-loaded trucks, and ore-carrying trains involving operation with very high axle pressure and greatly varying surrounding temperatures.

The object of the invention is to present a plant for the detection of imbalance in the wheels on ve¬ hicles running on rails, so that the wheel imbal¬ ance can be detected by means of a rail-mounted measuring system.

This is achieved by configuring the plant according to the invention as characterized in claim 1. All of the wheels which pass the measuring section can be automatically detected for imbalances upon pass- ing at normal running speed. After the passing of the train, on the basis of the signal values meas¬ ured it is possible to evaluate, automatically or manually, whether one or more of the wheels have unacceptable imbalance. For the identification of the individual carriages, a normal carriage ident¬ ification system can be used, e.g. a microwave-based system with the antenna placed at or in the vicinity of the measuring section, so that all of the car¬ riage numbers are read, and in such a manner that the carriage identification system is connected to the data processing unit in the plant, so that all sets of data can be provided with carriage numbers and possibly with wheel identification.

By configuring the plant according to the invention as characterized in claim 2, it is possible in an effective manner to separate the displacement or acceleration signals, which are characteristic for possible mass imbalances, from those signals which stem from other irregularities in the rails or the • wheels, e.g. shock effects arising from flats on the running surfaces of the wheels or rail vibra¬ tions, from the small, evenly-distributed irregu¬ larities on the surface of the rail and the wheels, so-called roller noise.

The separation can be optimized if filters with adjustable limiting frequencies are used, where the limiting frequencies are adjusted in accordance with

the speed of the train and possibly also the .size of the wheels. The speed is known, in that this is calculated on the basis of the signals from the transducers and/or the wheel detectors, and the wheel size can be measured, e.g. during the passing of the train, or can be found by means of the car¬ riage identification"system.

The plant's measuring system is preferably arranged as disclosed and characterized in claim 3, hereby providing a measuring window under each wheel, in that the transducers are coupled into a suitable window around the wheel, gradually as this passes the individual transducer. The possibility is here- by provided of distinguishing the individual wheels from one another, hereby enabling not only the iden¬ tification of which carriage or which bogie has an imbalanced wheel, but also the identification of the wheel which is out of balance. Furthermore, this provides the possibility of carrying out meas¬ urements on wheels of different diameters with the same plant. The on/off change of the transducers is set in accordance with the actual speed at which each wheel passes, so that it is also possible to measure imbalances during acceleration and braking.

It has proved to be advantageous to arrange the plant according to the invention as characterized in claim 4. In practice, the transducers can be disposed midway between successive sleepers, and the measuring window thus has a breadth which cor¬ responds to the distance between two successive sleepers or transducers. The transducers can also be disposed in every second space between the

sleepers or, if it is desired to use a broader meas¬ uring window, in every third space. Even though the transducers are disposed at fixed distances on the rail, e.g. corresponding to the distance between the sleepers, one can also purely electrically change the breadth of the measuring window by suitable con¬ trol of the coupling and decoupling of the trans¬ ducers.

The plant according to the invention can be config¬ ured as characterized in claim 5. In practice, one will of course carry out measurements on both of the rails in the rail element, and the plant can be ar¬ ranged with several wheel detectors, so that the plant senses the running direction before the- wheels reach into the actual measuring area. The wheel detectors can also be used for running meas¬ urement of the horizontal speed of the wheels, so that the plant is constantly aware of the precise speed during the course of the measurement. The wheel detectors are also used for the starting and partial closing-down of the plant, so that it meas¬ ures only when a train is passing the measuring sec¬ tion. Finally, the wheel detectors can also be used for detecting the size of the wheels.

The actual measuring procedure in the plant is pref¬ erably arranged as characterized in claims 6, 7 or 8, hereby eliminating the contribution from the sta- tic wheel load, i.e. the contribution from that part of the transducer signal which stems from the weight of the train, including its load. The resultant im¬ balance index is a measure of the magnitude of the imbalance, and can therefore form the basis for an

automatic or manual qualitative evaluation of how quickly the relevant wheel or the relevant carriage should be taken out of service for repair or mach¬ ining of the wheels.

By configuring the plant according to the invention as characterized in claim 9, it can be used to dis¬ tinguish between whether the imbalance signal stems from the wheel in the middle of the measuring win- dow, as described above, or whether it stems from a neighbouring wheel. This embodiment is particularly suitable for plants where it is desirable to be able to identify precisely that wheel which is out of balance. In some cases it will be satisfactory to be able to identify the bogie or possibly the complete carriage so that it can be removed from the train. This embodiment is also well suited for measurements on carriage constructions with closely-set neigh¬ bouring wheels, e.g. bogies with 4 axles.

A preferred embodiment of the plant according to the invention is configured as characterized in claim 10. The plant achieved hereby is based on the fact that not only does the whole of the measuring se- quence take place at the measuring section, but also that the whole sequence of calculation is effected at or in the vicinity of the measuring section. Those signals which are relayed further-- from the plant are the sets of data which are representative of the condition of the wheels with regard to im¬ balance, and one can naturally also arrange the plant in such a manner that only the data for wheels or carriages with an imbalance index above a certain magnitude are relayed further. This is of

great practical value, the reason being that one generally only has access to the use of a normal telephone line as transmission channel with the en¬ suing capacity thereof.

The invention will now be described in closer de¬ tail with reference to the drawing, where

fig. 1 shows the plant's transducers placed on a rail in the measuring section,

fig. 2 shows the plant as a whole in the form of a schematic diagram of the measuring sec¬ tion seen from above,

fig. 3 is a block diagram showing the entire plant according to a preferred embodiment,

fig. 4 shows frequency characteristics for vertical acceleration of a given point on a rail upon passage of a train wheel,

fig. 5 shows the measurement-signal contribution at pure imbalance without rail suppression.

fig. 6 shows time-displaced acceleration signals from wheels of different types, and

fig. 7A and B shows imbalance signals at a broad measuring window.

In figs. 1 and 2 of the drawing is seen that part of the plant which is mounted directly on the rails 2 of the measuring section 1. On each rail 2 in the

shown example there" are placed seven transducers3, which in the embodiment discussed hereafter are ac¬ celerometers. The accelerometers 3 are placed mid¬ way between neighbouring ^' sleepers 4, and thus the actual measuring section assumes a length of seven consecutive intervals between sleepers. At each end of the measuring section there are placed a number of wheel sensors 5. In the example shown there are three wheel sensors on each rail. This positioning of the accelerometers and the wheel sensors is only an example of how the plant can be arranged. To be able to measure as desired, the measuring section must at least have a length corresponding to the wheel circumference of a passing wheel, but can nat- urally be longer.

Under each accelerometer in fig. 1, it is also indi¬ cated whether it is coupled or decoupled, depending on where the wheel 7 on the bogie 6 is situated at the moment. The coupling and decoupling of the ac¬ celerometers occurs synchronously with the passing wheel, and in such a manner that each accelerometer is coupled in a suitable measuring window around the wheel. In the example shown, the width of the measuring window corresponds to the sleeper distance or the distance between the accelerometers.

All accelerometers 3 and all wheel sensors 5 are connected to the electronic signal processing unit 8, which in fig. 2 is shown sketched at the side of the rail element. From the electronic unit 8, which also comprises the necessary programs for the con¬ trol of the plant and the calculations herein, the results are sent via a transmission line 9, e.g. a

telephone line, to a central computer or to other surveillance equipment.

In fig. 3, the whole of the plant according to the invention is seen in more detail. Via the amplifier group 10,11,12,13, the accelerometers 3 and the wheel detectors 5 are coupled to a multiplex analog/ digital converter, in that the signals from the ac¬ celerometers are first filtered by means of the groups of filters 14 and 15. The amplifier groups and the filters are preferably arranged as shown, \so that there is an amplifier for each accelerometer or each wheel sensor, and in such a way that there is a filter for each accelerometer. The filters in the filter groups 14 and 15 are discussed in more detail later. The digital signals from the analog/digital converter 16 are sent via a databus 17 to the com¬ puter 18. Th s comprises a number of data collec¬ tion buffers and a program-controlled computer with time control which controls the whole of the plant and synchronizes the measuring and calculation pro- • cedure, which is discussed in more detail later.

The function of the filters in the filter groups 14 and 15 is to filter out some signal contributions from the accelerometers 3 which do not concern im¬ balance. In fig. 4 is seen the frequency character¬ istics of the types of signals which the accelero¬ meter will emit upon passage of a train wheel.

Fig. 4A shows the signal from the deflection of the rail arising from the static wheel load. The signal here is a very low-frequency, sinusoidal signal, where the frequency depends on the train's speed.

Fig. 4B shows the signal from additional or reduced deflection of the rail, arising as a result of an unbalanced wheel. The frequency is also low, norm¬ ally below 200Hz, and is dependent upon the train's speed and the wheel diameter.

Fig. 4C shows a signal from the shock effects which arises as a result of a flat on the roller surface of the wheel. The frequency of a signal from a wheel flat is broad-banded and lies in the range DC-3000 Hz.

Fig. 4D shows the signal for rail vibrations from small irregularities evenly distributed on the sur- face of the rail and the wheel, so-called roller noise. The frequency of this signal lies relatively high.

By using a suitable low-frequency filter for each accelerometer, e.g. a bandpass filter with the fre¬ quency range 1-2000Hz, preferably 1-100Hz, the con¬ tributions shown in figs. 4C and 4D are eliminated, in that a subsequent level evaluation is carried out in the computer. The separation can possibly be improved by using filters with variable limiting frequency, where the limiting frequency is adjusted on the basis of knowledge of the speed of the train and possibly on the basis of the diameter of the wheels.

Remaining are the signals for imbalance and the static wheel load, which are separated hereafter.

If the track at the measuring section is uniform

with equal distance between the sleepers and the same ballast, all seven signals from the seven ac¬ celerometers will be identical if they are passed only by perfectly round wheels without imbalance. In this case, the time-base course of the mean value of these seven signals will be equal to the time-base course of the individual signal, and the dispersion will ideally be equal to zero. When the accelerometers are passed by a wheel which has an imbalance, the signals from the individual acceler¬ ometers will not be the same. The fact that there is worked with, for example, seven accelerometers, makes it possible to calculate the mean value and the dispersion.

The above is based on the fact that when the wheel runs in on that piece of the rail which is covered by the individual accelerometer, the accelerometer will be forced downwards. When the wheel is oppo- site the accelerometer, the piece of rail will have reached its maximum deflection. Thereafter, the downwards deflection will again be reduced concur¬ rently with the movement of the wheel away from the middle of the piece of rail. The deflection of the rail piece will depend on the total dynamic force on the rail piece, and it will be mainly that dy¬ namic force which arises from the static wheel load and the force which is caused by a possible imbal¬ ance.

Seen over a passage sequence, the signal which ar¬ ises from that deflection which stems solely from the imbalance will appear as shown in fig. 5. In reality, however, the signal 25 in fig. 5 is super-

posed by a signal with a much lower frequency, namely the signal from the static wheel load, but this part of the signal is eliminated by the sub¬ sequent data processing.

In the column marked 26 in fig. 6 is seen the ac¬ celeration signal from all seven accelerometers for a perfect wheel and with a window corresponding to the distance between two sleepers. In column 27 are seen the signals for a wheel with a diameter of approx. 0.8 m and with an imbalance corresponding to the imbalance shown in fig. 5. In column 28 are seen the acceleration signals for a wheel with a diameter of approx. 0.4m and with an imbalance of the same magnitude as shown in column 27. The time base of the signals in the columns is displaced from one to six window breadths, so that they are superimposed on one another as shown at the bot¬ tom of the columns. In this example, the breadth of the window is the same as the distance between two neighbouring accelerometers. As a starting point for the displacement, the centre of the measuring window is used, which corresponds to the position of the accelerometer, so that the centre of all the measuring windows which are to be compared are brought over each other.

The signals in the column 26 for a perfect wheel are thus those stemming solely from the static wheel load, while the signals in columns 27 and 28 show imbalance signals without superposition from the static wheel load.

A calculation of the dispersion for the signals in

column .26 gives a dispersion of zero, and for col¬ umns 27 and 28 a dispersion which is different from zero. A calculation can also be made of an imbalance index, e.g. by integration of the hatched areas, whereby one arrives at an index as a simple number for each wheel. A perfect wheel will have an imbal¬ ance index of zero, and the greater the imbalance the greater will be the imbalance index.

For the sake of convenience, the above explanation has hitherto only dealt with measuring windows cor¬ responding to the distance between two sleepers.

Since the deflection of the rail system extends far on each side of the actual wheel, a possible imbal¬ ance of a closely-set neighbouring wheel will be able to exercise significant influence on the meas¬ urement of the actual wheel.

By introducing a broader window, e.g. seven times the distance between sleepers (the transducer dist¬ ance) , it will at the same time be possible to ev¬ aluate, without any change in what has been explain¬ ed earlier, whether the imbalance signal should pos- sibly stem mainly from a neighbouring wheel..

If the calculated dispersion (calculated as des¬ cribed earlier) is unsymmetrical, so that it in¬ creases and diminishes respectively across the now broad measuring window, see fig. 7B, there is a great probability that the imbalance stems from a neighbouring wheel.

On the other hand, if the dispersion is symmetrical

and does not significantly diminish or increase, there is a great probability that the imbalance stems from the measured wheel, see fig. 7A.

The distance between the vertical lines in figs. 7A and 7B corresponds to the passage of the space be¬ tween two sleepers (distance between the transduc¬ ers) .

The measuring plant according to the invention is described above as an independent plant. Without any deviation from the spirit of the invention, it is of course possible, and often expedient, to build the measuring plant according to the invention to- gether with other measuring plants, e.g. with a measuring plant for the measuring of wheel flats, where the same transducers/accelerometers and pos¬ sibly several other parts are used for both meas¬ uring procedures.