Field of the invention

[0001] The present invention generally relates to the field of rail transportation, and more particularly, to determining the angle of attack of a rail vehicle wheelset, and the hunting condition of the wheelset, bogie and/or vehicle. The present invention may also apply to determining a condition of a wheel of a rail vehicle and/or a rail of a railway track.

Background of the invention

[0002] In railway operation, monitoring angle of attack (AoA) and hunting performance of rail vehicles is important to ensure safe operation of railway vehicles, ride comfort, reduction in component wear and reduced maintenance costs. AoA is a particularly critical factor for assessing a rail vehicle’s curving performance. AoA is defined as the angle between wheels of a wheelset and the rails. A wheelset having a small AoA will result in lower wheel/rail forces and therefore lower component wear. In contrast, a wheelset having a large AoA will result in larger wheel/rail forces, thereby causing excessive wheel flange and rail side wear. Such increased wear ultimately results in higher maintenance costs. Wheelsets having too large an AoA will generate high lateral wheel/rail forces that may result in derailment and/or excessively noisy wheel squeal events (over a maximum noise level of LAFmax 115dBA at 2.5 metres from track).

[0003] Hunting is a critical factor for assessing a rail vehicle’s performance on tangent sections of track (i.e. straight track) and at high speed. Hunting is a condition in which a wheelset, bogie or rail vehicle swings from side to side between rails while the train operates at speed on a tangent section. Inadequate bogie design, high conicity of wheel profiles due to design or wear, worn rail profiles and other reasons contribute to hunting. Hunting results in rapid wear of vehicle components and rails, damage to transported goods, serious ride comfort issues, and possibly derailment.

[0004] There have been various approaches adopted in the past to measure AoA and hunting. In one approach, a wayside optical system comprising a laser beam and a camera, or multiple laser beams and multiple cameras, are used to measure AoA using the principle of optical triangulation. In addition to the optical sensors and detectors, wheel sensors are required for triggering the measurement or for measuring speed. The optical systems are sensitive to lighting conditions, wheel rim surface contamination (e.g. having debris, dirt, or the like), and wheel/rail vibration, and may result in inconsistent or inaccurate measurements under these conditions. In another approach, proximity sensors may be deployed, one to each rail, to measure the duration and relative timing of the passing wheels, to enable estimation of the AoA. As the output of the wheel sensors are binary with only two states (ON or OFF), the duration and relative timing determined from the wheel sensors are sensitive to the wheel diameter, lateral position, and train speed. Any variation in response to these parameters can result in an inaccurate or incorrect AoA measurement.

[0005] In yet another approach, strain gauges, as a variant of proximity sensors, are mounted to the rail to measure the vertical and lateral forces. The ratio of the lateral force to the vertical force is indicative of wheelset alignment, and provides a reasonable estimation of AoA when the AoA is small. However, when the AoA increases, the wheel/rail interface friction may saturate, and the ratio of the lateral force to the vertical force may approach a constant despite an increasing AoA. The installation of strain gauges on a track typically requires grinding of the rail, and replacement of concrete sleepers.

[0006] In view of the above, there is a desire to provide improved systems and methods for determining the AoA of a wheelset of a rail vehicle, or at least workable alternatives. There is also a desire to provide improved systems and methods for determining a condition of a wheel and/or rail, or at least workable alternatives.

[0007] Reference to any prior art in the specification is not an acknowledgment or suggestion that this prior art forms part of the common general knowledge in any jurisdiction or that this prior art could reasonably be expected to be understood, regarded as relevant, and/or combined with other pieces of prior art by a skilled person in the art. Summary of the invention

[0008] In a first aspect, the present invention provides a system for measuring angle of attack of a wheelset of a railway vehicle travelling on a section of railway track, the system including: a first sensor configured to be mounted on a first rail of the railway track and to output a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset; a second sensor configured to be mounted on a second rail of the railway track and to output a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset, wherein the first and second sensors are substantially aligned along an axis perpendicular to the railway track; a data acquirer configured to receive the first and second continuous analog wheel sensor signals from the first and second sensors and convert the continuous analog wheel sensor signals to digital signals, the digital signals including peak indicia representative of the times of arrival of the centres of the first and second wheels; and a processor configured to determine the angle of attack of the wheelset from the digital signals based on the difference in times of arrival of the centres of the first and second wheels.

[0009] Advantageously, the first and second sensors are able to capture the complete movement of the wheels (from approaching, arriving and departing) to allow accurate determination of the time of arrival of each wheel centre. This enables accurate determination of the angle of attack of the wheelset.

[0010] Accurate determination of the angle of attack of a wheelset will better inform when a wheelset of a bogie requires maintenance or replacement. For example, some prior art systems may have a tendency to underestimate the angle of attack of a particular wheelset. This may result in continued service use of a deteriorating bogie, causing possible safety issues, excessive wheel/rail wear, and/or severe wheel squeal.

[0011] The angle of attack may be defined as the angle between a longitudinal axis of a wheelset axle and an imaginary axis perpendicular to the tangent of the railway track at the point of engagement between the wheel and the track.

[0012] Preferably, the first and second sensors are inductive sensors. However, it will be appreciated by a person skilled in the art that other types of sensors may be used so long as they are able to output a continuous analog wheel sensor signal sensing the approaching, arriving and departing of the wheels.

[0013] In an embodiment, the processor is configured to estimate from the peak indicia of the digital signals, the times of arrival of the centres of the first and second wheels using a peak detection algorithm or method. This can be achieved by a method of fitting a line or curve to the digital signals to identify one or more peaks in the signal, with the peaks in the digitals signals coinciding with the arrival of the respective wheel centre. Preferably, the peaks in the digital signals are estimated using a peak detection algorithm. The peak detection algorithm may be in the form of a quadratic interpolation. However, other mathematical peak detection and curve fitting algorithms can also be used such as those using higher order polynomial functions, Gaussian functions, or any other peak detection functions and/or methods for curve fitting and interpolation that could routinely be conceived by a person skilled in the art.

[0014] In an embodiment, the continuous analog wheel sensor signals from the first and second sensors are converted to respective digital signals at a high sampling rate typically based on the operational conditions of the rail vehicle. In one example, the sampling rate may be equal to or greater than 25.6 kHz (25,600 samples per second). For hunting detection (as will be explained in more detail below) a higher sampling rate may be used for higher rail vehicle speeds. Advantageously, selecting a suitable sampling rate based on the operational conditions of the rail vehicle can result in the digitised signals retaining the detail of their respective analog signals when being further processed.

[0015] In an embodiment, the processor is configured to offset the digital signals so that the outputs of the first and second sensor are substantially equal in value when a wheel is not present. For example, the output of the first and second sensors may be offset to about 0 volts (if using voltage), or any other identical voltage level or datum point. This is particularly beneficial to eliminate issues associated with variation in wheel sensor settings and installation, which may otherwise cause the sensors to output signals of different level for effectively the same measurement. [0016] Preferably, the processor is configured to apply a predetermined threshold value to the commonly offset digital signals. Application of such a threshold enables determination of the total number of wheel pulses present in the digital signals, with this number corresponding to the total number of wheels detected to have passed the first and second sensors. The predetermined threshold value may be set to be slightly less than the offset output value (typically a predetermined voltage or current threshold value) corresponding to when a wheel is not present. Advantageously, this substantially reduces or avoids the occurrence of missing a passing wheel. This also assists in terms of monitoring and maintenance as wheel pulses in the digital signal can be associated with the specific wheel/wheelset detected by the sensors.

[0017] In an embodiment, the processor is configured to filter the digital signals in order to substantially remove distortion caused by wheel and/or rail vibration and/or to reduce any estimation errors if present. The presence of severe wheel/rail vibration distorts the digital signals in a way that can result in false detection of wheel pulses and prevent peaks from being detected accurately. In one example, low-pass filtering is utilised to smooth the digital signals prior to any wheel pulse and/or peak detection. It will be appreciated that analog filtering (e.g. low-pass filtering) may also be applied to the analog signals before they are digitised.

[0018] In an embodiment, the processor is configured to calibrate the determined angle of attack using one or more other angles of attack determined at the section of the railway track. In one embodiment, the processor is configured to calibrate the determined angle of attack using one or more angles of attack determined from one or more trailing wheelsets. This is particularly beneficial in compensating for any alignment error between the first and second sensor when the system is installed on a curved section of the railway track because it is generally well understood in the art that trailing wheelsets generally exhibit near zero angle of attack on curved sections. This information can be utilised by the processor to calibrate the determined angle of attack accordingly. For example, an average of the one or more other angles of attack can be used for the calibrating. In another embodiment, the processor is configured to calibrate the determined angle of attack by using one or more angles of attack determined from a combination of one or more trailing and leading wheelsets at the section of the railway track. This is particularly beneficial in compensating for any alignment error between the first and second sensor when the system is installed on a tangent section of the railway track. It will appreciated that in some embodiments, the calibration can be done on a rail vehicle-by-rail vehicle basis, based on all determined angles of attack for a plurality of rail vehicles captured in the digital signals, and/or by only using determined angles of attack of wheelsets considered to be of good steering order.

[0019] In an embodiment, the system includes a third sensor configured to be mounted on the second rail, the second and third sensors spaced apart by a predetermined distance and configured to measure speed of the second wheel by using a difference in time of arrival of the second wheel passing the second and third sensors. Preferably, the measured speed of the second wheel is used when determining the angle of attack of the wheelset. The measured speed may provide an accurate representation of the speed of the rail vehicle.

[0020] In an alternative to the above embodiment, the second sensor includes at least two integral sub-systems spaced apart by a predetermined distance, thereby enabling the second sensor to measure speed of the second wheel by using a difference in time of arrival of the second wheel passing the two sub-systems. Preferably, the first sensor may also include at least two integral sub-systems, thereby enabling the first sensor to measure speed of the first wheel by using a difference in time of arrival of the first wheel passing the two sub-systems. The processor may then have the option of deciding which or both of the measured wheel speeds to use when determining the angle of attack of the wheelset.

[0021] In an embodiment, the data acquirer and the processor are part of an integral computing unit. Alternatively, the system may also include a separate computing system. The computing system preferably includes the processor, wherein the computing system is configured to store and process the digital signals. The computing system may include a communication module to transmit the digital signals to a remote server, whereby a remote processing unit can access the digital signals for processing.

[0022] In a second aspect, the present invention provides a method for determining angle of attack of a wheelset of a railway vehicle travelling over a section of a railway track, the method including: acquiring a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset; acquiring a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset; converting the first and second continuous analog wheel sensor signals to digital signals, the digital signals including peak indicia representative of the times of arrival of the centres of the first and second wheels; and determining the angle of attack of the wheelset from the digital signals based on the difference in the times of arrival of the centres of the first and second wheels.

[0023] In an embodiment, the method further includes estimating from the peak indicia of the digital signals, the times of arrival of the centres of the first and second wheels using a peak detection method or algorithm. This can be achieved by a method of fitting a line or curve to the digital signals to identify one or more peaks in the signal, with the peaks in the digitals signals coinciding with the arrival of the respective wheel centre. Preferably, the peaks in the digital signals are estimated using a peak detection algorithm. The peak detection algorithm may be in the form of a quadratic interpolation. However, other mathematical peak detection and curve fitting algorithms can also be used such as those using higher order polynomial functions, Gaussian functions, or any other peak detection functions and/or methods for curve fitting and interpolation that would routinely be conceived by a person skilled in the art.

[0024] In an embodiment, the continuous analog wheel sensor signals are converted to digital signals at a high sampling rate typically based on the operational conditions of the rail vehicle. For hunting detection (as will be explained in more detail below) a higher sampling rate may be used for higher rail vehicle speeds.

[0025] In an embodiment, the method further includes offsetting the digital signals so that the first and second continuous analog wheel sensor signals are substantially equal in value when a wheel is not present. For example, the first and second continuous analog wheel sensor signals may be offset to about 0 volts (if using voltage), or any other identical voltage level or datum.

[0026] In an embodiment, the method further includes applying a predetermined threshold value to the commonly offset digital signals, and determining a number of wheel pulses detected in the first and second continuous analog wheel sensor signals. This will correspond to the number of wheels detected in the first and second continuous analog wheel sensor signals. The predetermined threshold value may be set at slightly less than the offset output value corresponding to when a wheel is not present.

[0027] In an embodiment, the method further includes filtering the digital signals in order to substantially remove or reduce distortion caused by wheel and/or rail vibration and/or to reduce any estimation errors if present. In one example, low-pass filtering is utilised to smooth the digital signals prior to any wheel pulse and peak detection. It will be appreciated that analog filtering (e.g. low-pass filtering) may also be applied to the analog signals before they are digitised.

[0028] In an embodiment, the method further includes calibrating the determined angle of attack using one or more other angles of attack determined at the section of the railway track. For example, an average of the one or more other angles of attack can be used for the calibrating. In one embodiment, the calibrating includes using one or more angles of attack determined from one or more trailing wheelsets (e.g. for a curved section of the railway track). In another embodiment, the calibrating includes using one or more angles of attack determined from a combination of one or more trailing and leading wheelsets at the section of the railway track (e.g. for a tangent section of track). It will appreciated that in some embodiments, the calibration can be done on a rail vehicle by rail vehicle basis, be based on all determined angles of attack captured in the digital signals, and/or by only using determined angles of attack of wheelsets considered to be of good steering order.

[0029] In an embodiment, the method further includes acquiring a third wheel sensor signal sensing the arrival of the second wheel of the wheelset, determining speed of the second wheel based on the second and third wheel sensor signals, and using the speed of the second wheel to determine the angle of attack of the wheelset

[0030] In an embodiment, the method includes storing and processing the digital signals, wherein the angle of attack of the wheelset is determined from the processing. The method may further include communicating the digital data to a remote server, whereby the data can be accessed remotely for processing.

[0031] In a third aspect, the present invention provides a system for determining whether a wheelset of a railway vehicle travelling on a section of railway track is hunting, the system including: a first sensor pair configured to be mounted on opposed first and second rails of the railway track and output continuous analog wheel sensor signals sensing the approaching, arriving and departing of a first and second wheels of the wheelset, the first sensor pair substantially aligned along an axis perpendicular to the railway track at a first location; a second sensor pair configured to be mounted on the opposed first and second rails of the railway track and output continuous analog wheel sensor signals sensing the approaching, arriving and departing of the first and second wheels of the wheelset, the second sensor pair substantially aligned along an axis perpendicular to the railway track at a second location; a data acquirer configured to receive the continuous analog wheel sensor signals from the first and second sensor pairs and convert the continuous analog wheel sensor signals to digital signals, the digital signal including peak indicia representative of the times of arrival of the centres of the first and second wheels at the first and second locations respectively; a processor configured to determine the angle of attack of the wheelset at each of the first and second locations from the digital signals based on the difference in times of arrival of the centres of the first and second wheels at the first and second locations respectively, and thereby determine whether the wheelset is hunting based on the difference in angle of attack at the first and second locations.

[0032] In a fourth aspect, the present invention provides a method for determining whether a wheelset of a railway vehicle travelling on a section of railway track is hunting, the method including: acquiring a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset at a first location; acquiring a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset at the first location; acquiring a third continuous analog wheel sensor signal sensing the approaching, arriving and departing of the first wheel of the wheelset at a second location; acquiring a fourth continuous analog wheel sensor signal sensing the approaching, arriving and departing of the second wheel of the wheelset at the second location; converting the first, second, third and fourth continuous analog wheel sensor signals to digital signals, the digital signals including peak indicia representative of the times of arrival of the centres of the first and second wheels at the first and second locations respectively; and determining the angle of attack of the wheelset at each of the first and second locations from the digital signals based on the difference in times of arrival of the centres of the first and second wheels at the first and second locations respectively, and thereby determine whether the wheelset is hunting based on the difference in angle of attack at the first and second locations.

[0033] In a fifth aspect, the present invention provides a system for measuring angle of attack of a wheelset of a railway vehicle travelling on a section of railway track, the system including: a first sensor configured to be mounted on a first rail of the railway track and to output a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset; a second sensor configured to be mounted on a second rail of the railway track and to output a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset, wherein the first and second sensors are substantially aligned along an axis perpendicular to the railway track; a signal processing system configured to receive the first and second continuous analog wheel sensor signals from the first and second sensors and identify peaks representative of the times of arrival of the centres of the first and second wheels and to determine the angle of attack of the wheelset from the signals based on the difference in times of arrival of the centres of the first and second wheels.

[0034] The signal processing system may include analog circuitry for including an analog filter.

[0035] In a sixth aspect, the present invention provides a method for determining angle of attack of a wheelset of a railway vehicle travelling over a section of a railway track, the method including: acquiring a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset; acquiring a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset; identifying peaks representative of the times of arrival of the centres of the first and second wheels from the signals; and determining the angle of attack of the wheelset from the signals based on the difference in the times of arrival of the centres of the first and second wheels.

[0036] In a seventh aspect, the present invention provides a system for determining a condition of a wheelset of a railway vehicle and/or a rail of a railway track, the system including: a first sensor configured to be mounted on a first rail of the railway track and to output a continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel of the wheelset; a second sensor configured to be mounted on a second rail of the railway track and to output a continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel of the wheelset; and a signal processing system configured to: receive the continuous analog wheel sensor signals from the sensors; convert the continuous analog wheel sensor signals into digital signals; perform frequency analysis to identify frequency components in the digital signals; and determine the condition of the wheelset and/or rail.

[0037] In one embodiment, the condition of the wheelset and/or rail is a corrugation and/or a wheel squeal or vibration condition.

[0038] In an eighth aspect, the present invention provides a method for determining a condition of a wheelset of a railway vehicle and/or a rail of a railway track, the method including: acquiring a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel travelling on the rail; acquiring a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a second wheel travelling on the rail; converting the continuous analog wheel sensor signals into digital signals; identifying frequency components in the digital signals; and analysing the frequency components to determine the condition of the wheelset and/or rail.

[0039] In a ninth aspect, the present invention provides a method for determining a condition of a wheelset of a railway vehicle and/or a rail of a railway track, the method including: acquiring a first continuous analog wheel sensor signal sensing the approaching, arriving and departing of a first wheel travelling on a first rail; acquiring a second continuous analog wheel sensor signal sensing the approaching, arriving and departing of a wheel travelling on a second rail; converting the continuous analog wheel sensor signals into digital signals; identifying high frequency components in the digital signals; analysing the high frequency components to determine at least one vibration- related indicator of wheelset and/or rail condition; filtering the digital signals to remove the high frequency components; identifying peaks representative of the times of arrival of the centres of the first and second wheels; and determining the angle of attack of the wheelset from the filtered signals based on the difference in the times of arrival of the centres of the first and second wheels.

[0040] In an embodiment, the method further includes analysing the angle of attack of the wheelset and the corresponding vibration-related indicator(s) to determine the condition of the wheelset and/or the rail.

[0041] It will be appreciated that features disclosed with respect to any of the above aspects of the invention may also apply to any other aspect of the invention. For example, in an embodiment where the continuous analog wheel sensor signals are converted to digital signals, an unfiltered part of the digital signal can be used to determine the condition of the wheelset and/or rail, and the filtered part of the digital signal can be used to determine the angle of attack of the wheelset.

[0042] As used herein, except where the context requires otherwise, the term "comprise" and variations of the term, such as "comprising", "comprises" and "comprised", are not intended to exclude further additives, components, integers or steps.

[0043] Further aspects of the present invention and further embodiments of the aspects described in the preceding paragraphs will become apparent from the following description, given by way of example and with reference to the accompanying drawings.

Brief description of the drawings

[0044] Figure 1 illustrates a definition of angle of attack with reference to a wheelset travelling on a railway track;

[0045] Figure 2 illustrates a schematic top plan view of a system for determining angle of attack of a wheelset of a railway vehicle in accordance with one embodiment of the invention; [0046] Figure 3 illustrates a rear view of one possible installation of a pair of wheel sensors on adjacent rails of the railway track;

[0047] Figure 4 illustrates a continuous analog wheel sensor signal from an inductive wheel sensor used in at least one embodiment of the present invention;

[0048] Figure 5 is a flow diagram illustrating the signal processing steps used for determining the angle of attack of at least one embodiment of the present invention;

[0049] Figure 6 illustrates an example of offsetting wheel pulses to achieve reliable wheel pulse detection in accordance with an embodiment of the present invention;

[0050] Figure 7a-b illustrates examples of wheel sensor signals prior and after low- pass filtering in accordance with an embodiment of the present invention;

[0051] Figure 8 illustrates a peak detection method using quadratic fitting and interpolation to detect exact time of arrival of wheel centres of the rail vehicle;

[0052] Figure 9 illustrates another definition of angle of attack with reference to a wheelset travelling on a railway track;

[0053] Figure 10 is a block diagram of a computer processing system configurable to perform various features of the present invention;

[0054] Figure 11 illustrates an example of calculating wheel flange distance;

[0055] Figure 12 illustrates a measured difference in angle of attack (relative to a theoretical difference) for a passenger train with 32 axle passes;

[0056] Figure 13 illustrates a measured difference in angle of attack (relative to a theoretical difference) for a freight train with 266 axle passes;

[0057] Figure 14 provides a histogram for measured angle of attack difference for 1,017,718 axle passes;

[0058] Figure 15 is a schematic top plan view of a hunting detection system; [0059] Figure 16 is a flow diagram illustrating the signal processing steps used for determining a condition of a wheel/rail of at least one embodiment of the present invention; and

[0060] Figure 17a-c illustrating wheel sensor signals prior and after undergoing frequency analysis.

Detailed description of the embodiments

[0061] Reference is made to Figure 1 , which illustrates a wheelset of a rail vehicle. The wheelset includes a first wheel 2 and a second wheel 4 rigidly connected to a solid axle, with the first wheel 2 running on a first rail 6 of a railway track and the second wheel 4 running on a second rail 8 of the railway track. The direction of travel is right to left as indicated by the arrow in Figure 1.

[0062] It will be appreciated from Figure 1 that the wheels 2, 4 are not perfectly aligned with the rails 6, 8, but instead an angle exists between the plane of each respective wheel and an imaginary axis tangent to the rail. This angle is known as the angle of attack a, which generally refers to the yaw angle of the wheelset relative to the rail resulting from the inability of the wheelset to perfectly align with the rail tangent. This is generally most noticeable when the wheelset of a rail vehicle is negotiating a curved section of track.

[0063] When an angle of attack is present between the wheelset and the rail, the wheelset tends to “run” into one rail, resulting in wheel flange wear, rail side wear and increased wheel squeal (particularly when passing over curved sections of track). For this reason, it is important to accurately determine the angle of attack of a passing wheelset in order to ascertain the overall steering performance of the rail vehicle, which generally include a plurality of wagons each fitted with one or more bogies as is understood in the art.

[0064] Reference is now made to Figure 2, which illustrates an angle of attack determination system 100. System 100 includes a first wheel sensor 10 installed on the first rail 6, and a second and third wheel sensors 12, 13 installed on the second rail 8. It will be appreciated from Figure 2 that the first wheel sensor 10 on first rail 6 and the second wheel sensor 12 on the second rail 8 are installed substantially along an imaginary line perpendicular to the longitudinal axis of rails 6, 8. This sensor pair is used in the system 100 for measuring the angle of attack of a passing wheelset. The third wheel sensor 13 is installed on the second rail 8 at a predetermined distance from the second wheel sensor 12. The second and third sensors 12, 13 are used in system 100 to calculate the speed of the second wheel, and hence provide a good approximation for the speed of the rail vehicle. A person skilled in the art will appreciate that the third sensor 13 can be positioned either upstream or downstream of the second sensor 12. The speed of the rail vehicle is used as part of the angle of attack calculation, an example of which will be described in more detail below.

[0065] The wheel sensors 10, 12, 13 of this embodiment are inductive wheel sensors that output a continuous analog signal. It is to be understood that continuous in this sense refers to the sensors sensing the approaching, arriving and departing of a given wheel. The sensors do not have to always be powered on (can be toggled on and off as needed), but should be able to capture “continuous” information pertaining to the approaching, arriving and departing of a given wheel. Inductive wheel sensors have some advantages over other forms of detection means, as they are generally more robust when exposed to different environmental conditions or in the presence of dirt or debris. One example of suitable wheel sensors for this embodiment of the invention include the Frauscher RSR110 wheel sensor. A wheel sensor such as this particular Frauscher sensor has two built-in systems, which are used for speed and direction information. These two built-in systems have a known spacing (about 131 mm with tolerances of ±0.1 mm), enabling the one wheel sensor to accurately measure the speed of a passing wheel. However, a person skilled in the art will appreciate that other types of wheel sensors may be suitable.

[0066] The first and second wheel sensors 10, 12 may be installed above a sleeper or between sleepers, inside (track side) or outside (field side) of the rails, depending on the wheel sensors used (different manufacturers often have different installation specifications). With reference to Figure 3, wheel sensors 10, 12 are installed between sleepers and on the inside of the rails 6, 8. Each wheel sensor is secured to the rail using a mounting bracket 20 as is conventional in the art. A similar approach is preferably adopted for mounting third wheel sensor 13. [0067] Returning to Figure 2, the wheel sensors 10, 12, 13 are connected to a data acquisition unit 16 via signal cables 14. The data acquisition unit 16 is configured to receive the continuous analog wheel sensor signals output from wheel sensors 10, 12, 13 and convert the analog signals into digital signals at a suitable sampling rate based on the operational conditions of the rail vehicle using an A/D converter. For example, a sampling rate no less than 25.6KHz is preferred for measuring the angle of attack of the wheelset on a curved section of track for a rail vehicle travelling at a speed less than about 100 km/h to ensure the digital signals include sufficient data from the continuous analog wheel sensor signals. This ensures that the continuous analog wheel sensor signals capture data when wheel/rail vibration is at squealing frequencies, typically around 5000 Hz or more.

[0068] As shown in Figure 2, the data acquisition unit 16 is generally positioned adjacent the railway track, and is in wired communication with the wheel sensors. However, in alternative embodiments, the data acquisition unit 16 may communicate with the wheel sensors 10, 12, 13 wirelessly. The digital signals can be stored by the data acquisition unit 16 internally (in memory) or externally (e.g. on an external USB drive or using cloud storage). Remote access to the data acquisition unit 16 may also be available, for example, via an integrated 3G modem in the system. A suitable data acquisition unit for such a system is the CompactRIO system provided by National Instruments.

[0069] The acquired wheel sensor signals can be stored and processed in a computer processing system 500. One example of a computer processing system is described below with reference to Figure 10, which provides a block diagram of a computer processing system 500 configurable to implement embodiments and/or features described herein. System 500 is a general purpose computer processing system. It will be appreciated that Figure 10 does not illustrate all functional or physical components of a computer processing system. For example, no power supply or power supply interface has been depicted, however system 500 will either carry a power supply or be configured for connection to a power supply (or both). It will also be appreciated that the particular type of computer processing system will determine the appropriate hardware and architecture, and alternative computer processing systems suitable for implementing features of the present disclosure may have additional, alternative, or fewer components than those depicted.

[0070] Computer processing system 500 includes at least one processing unit 502 - for example a general or central processing unit, a graphics processing unit, or an alternative computational device). Computer processing system 500 may include a plurality of computer processing units. In some instances, where a computer processing system 500 is described as performing an operation or function all processing required to perform that operation or function will be performed by processing unit 502. In other instances, processing required to perform that operation or function may also be performed by remote processing devices accessible to and useable by (either in a shared or dedicated manner) system 500.

[0071] Through a communications bus 504, processing unit 502 is in data communication with one or more computer readable storage devices which store instructions and/or data for controlling operation of the processing system 500. In this example system 500 includes a system memory 506 (e.g. a BIOS), volatile memory 508 (e.g. random access memory such as one or more DRAM modules), and non-volatile (or non-transitory) memory 510 (e.g. one or more hard disk or solid state drives). Such memory devices may also be referred to as computer readable storage media.

[0072] System 500 also includes one or more interfaces, indicated generally by 512, via which system 500 interfaces with various devices and/or networks. Generally speaking, other devices may be integral with system 500, or may be separate. Where a device is separate from system 500, connection between the device and system 500 may be via wired or wireless hardware and communication protocols, and may be a direct or an indirect (e.g. networked) connection.

[0073] Wired connection with other devices/networks may be by any appropriate standard or proprietary hardware and connectivity protocols, for example Universal Serial Bus (USB), eSATA, Thunderbolt, Ethernet, HDMI, and/or any other wired connection hardware/connectivity protocol.

[0074] Wireless connection with other devices/networks may similarly be by any appropriate standard or proprietary hardware and communications protocols, for example infrared, BlueTooth, WiFi; near field communications (NFC); Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), long term evolution (LTE), code division multiple access (CDMA - and/or variants thereof), and/or any other wireless hardware/connectivity protocol.

[0075] Generally speaking, and depending on the particular system in question, devices to which system 500 connects - whether by wired or wireless means - include one or more input/output devices (indicated generally by input/output device interface 514). Input devices are used to input data into system 500 for processing by the processing unit 502. Output devices allow data to be output by system 500. Example input/output devices are described below, however it will be appreciated that not all computer processing systems will include all mentioned devices, and that additional and alternative devices to those mentioned may well be used.

[0076] For example, system 500 may include or connect to one or more input devices 514 by which information/data is input into (received by) system 500. Such input devices may include keyboards, mice, trackpads (and/or other touch/contact sensing devices, including touch screen displays), microphones, accelerometers, proximity sensors, GPS devices, touch sensors, and/or other input devices. System 500 may also include or connect to one or more output devices 514 controlled by system 500 to output information. Such output devices may include devices such as displays (e.g. cathode ray tube displays, liquid crystal displays, light emitting diode displays, plasma displays, touch screen displays), speakers, vibration modules, light emitting diodes/other lights, and other output devices. System 500 may also include or connect to devices which may act as both input and output devices, for example memory devices/computer readable media (e.g. hard drives, solid state drives, disk drives, compact flash cards, SD cards, and other memory/computer readable media devices) which system 500 can read data from and/or write data to, and touch screen displays which can both display (output) data and receive touch signals (input).

[0077] System 500 also includes one or more communications interfaces 516 for communication with a network. Via a communications interface 516, system 500 can communicate data to and receive data from networked devices, which may themselves be other computer processing systems (such as data acquisition unit 16). [0078] System 500 stores or has access to computer applications (also referred to as software or programs) - i.e. computer readable instructions and data which, when executed by the processing unit 502, configure system 500 to receive, process, and output data. Instructions and data can be stored on non-transitory computer readable medium accessible to system 500. For example, instructions and data may be stored on non-transitory memory 510. Instructions and data may be transmitted to/received by system 500 via a data signal in a transmission channel enabled (for example) by a wired or wireless network connection over interface such as 512.

[0079] Applications accessible to system 500 will typically include an operating system application such as Microsoft Windows®, Apple OSX, Apple IOS, Android, Unix, or Linux.

[0080] In some cases part or all of a given computer-implemented method will be performed by system 500 itself, while in other cases processing may be performed by other devices in data communication with system 500.

[0081] The flowcharts illustrated in the figures and described below define operations in particular orders to explain various features. In some cases, the operations described and illustrated may be able to be performed in a different order to that shown/described, one or more operations may be combined into a single operation, a single operation may be divided into multiple separate operations, and/or the function(s) achieved by one or more of the described/illustrated operations may be achieved by one or more alternative operations. Still further, the functionality/processing of a given flowchart operation could potentially be performed by different systems or applications.

[0082] Reference is now made to Figure 4, which illustrates an example of a continuous analog wheel sensor signal 40 captured from a passing wheel. The output of the wheel sensor remains constant when there is no wheel passing such as in regions 42, 46. The wheel sensor signal experiences rising, peaking and then falling, which corresponds to the approaching, arrival and departing of the wheel in region 44. The peak (or valley) 48 of the wheel sensor signal 40 corresponds to the exact time of arrival of the wheel centre, and is used to accurately determine the angle of attack in this embodiment. The part of the signal corresponding to the passing wheel will be referred to as a wheel pulse. It will be appreciated that the analog wheel sensor signal shown in Figure 4 is a voltage over time.

[0083] Figure 5 outlines the signal processing steps undertaken in order to accurately determine the angle of attack of a wheelset in this embodiment. These steps generally involve the following: data acquisition and A/D conversion 51, offsetting the wheel sensor signal 52, low pass filtering 53, wheel detecting 54, peak detection 55, speed and AoA estimation 56, and adjusting or calibration 57. It will be appreciated that these steps need not all be conducted in the precise order shown in Figure 5 and disclosed below. It will also be appreciated that whilst conducting all the steps may provide a more accurate angle of attack determination, each and every step need not be done. The signal processing steps and algorithms used can be performed using suitable software code, or suitable hardware such as a reconfigurable field-programmable gate array (FPGA) or an application specific IC (ASIC).

Data capturing and conversion

[0084] The continuous analog wheel sensor signals from wheel sensors 10, 12, 13 are sent via cables 14 to the data acquisition unit 16. Any commercial data acquisition system, which is capable of capturing all channels (at least three channels in this example) of analog wheel sensor data simultaneously and converting the analog data into a digitised signal at a suitable sampling rate can be used. In this embodiment, the acquired digitised signal is stored and processed in the computer processing system 500 at a sampling rate of at least 25.6kHz.

[0085] Whilst the embodiment described herein involves the continuous analog wheel sensor signals being converted to digital signals, it is to be appreciated that other forms of signal processing may be employed. For example, the following signal processing steps may be undertaken with the continuous analog wheel sensor signals without ever converting the continuous analog wheel sensor signals to digital signals. In another alternative, some of the following signal processing steps can be undertaken on the continuous analog wheel sensor signals prior to conversion into digital signals.

Offsetting wheel sensor signals [0086] The constant output of wheel sensors in the absence of a wheel passing may vary due to the variation in wheel sensor settings and installation. In the event that large discrepancies exist between the constant output of the wheel sensors 10, 12, 13 there is a possibility that wheel pulses may not be detected. This is best illustrated with reference to Figure 6. Whilst it is possible when processing the data to analyse each individual wheel pulse in order to determine whether a wheel passing is detected, this would be a cumbersome process. Instead, it is more efficient when processing the digital signal to apply a predetermined threshold to the entire digital signal. If the detected pulse exceeds the threshold, this indicates that a wheel has passed.

[0087] As shown in the top signals 62A, 62B and 62C of Figure 6, a simple threshold 63 does not detect all three wheel pulses from wheel sensors 10, 12, 13 due to the large offsets between the output of the wheel sensor 10, 12, 13, but only detects the lower two signals 62B and 62C. However, when the constant output of the wheel sensors 10, 12, 13 in the absence of the wheel passing are adjusted to the same level (as shown in the lower signals 64A, 64B and 64C of Figure 6), a simple threshold 63 can be used to detect all three wheel pulses reliably. This enables the number of wheels/wheelsets that pass the wheel sensors to be reliably counted. From this count, it would be possible to identify precisely which rail vehicle or wheelset is associated with each signal. This can be achieved by using time stamp information in combination with known rail vehicle schedules or through reading AEI tags on the rail vehicle. The information acquired here can be used to advise when maintenance of a bogie is required if it is determined to have an unacceptably high angle of attack (for e.g. some standards such as the Asset Standards Authority set an angle of attack limit of 15 mrad).

[0088] Whilst the embodiment described herein involves offsetting the converted digital signals, it is to be appreciated that the continuous analog wheel sensor signals may be subjected to this offsetting and thresholding (without ever converting the continuous analog wheel sensor signals to digital signals or before such conversion).

Filtering wheel sensor signals [0089] Since the wheel sensors 10, 12, 13 are attached to the rails 6, 8 of the railway track, the output of the wheel sensors are inevitably affected by wheel/rail vibration. In the presence of large wheel/rail vibration, the wheel sensor output can be significantly distorted and may impact the detection of wheel pulses and hence cause errors in determining angle of attack. The vibration ranges from the lower frequency range due to corrugation, typically from 50 Hz to 1000Hz as shown by signals 72 in Figure 7(a), to higher frequencies (greater than 1000Hz) due to a squealing wheel as shown by signals 74 in Figure 7(b). The presence of severe wheel/rail vibration distorts the wheel sensor signal in a way that can result in false detection of wheel pulses and prevent peaks (valleys) of the wheel pulse from being detected accurately.

[0090] Low-pass filtering can therefore be utilised to smooth the wheel sensor signal prior to wheel pulse and peak detection. The cut-off frequency can be optimised for a specific wheel/rail system and the operating conditions. For example, if the system 100 is installed on a curved section of track, the cut-off frequency may be configured to 50 Hz to filter out the vibrations including rail corrugation or squealing wheel induced vibrations. A different approach may be adopted for a tangent track section or for different types of rail vehicles (e.g. freight vs. passenger trains).

[0091] Whilst the embodiment described herein involves filtering the converted digital signals, it is to be appreciated that analog filtering of the continuous analog wheel sensor signals may instead be implemented with a suitable analog filter. The analog filtering may be implemented on hardware having suitable circuitry as is known in the art.

Detecting wheel pulses

[0092] After the wheel sensor signal is pre-processed, including offsetting and filtering, the individual wheel pulses corresponding to each wheel can be readily detected/identified using a simple threshold as shown in Figure 6 and explained above with regard to offsetting. The threshold is generally configured to be slightly less than the constant output value of the wheel signals in the absence of a wheel passing to ensure detection of all wheel pulses. [0093] In alternative embodiments, wheel pulses can be detected from the continuous analog wheel sensor signals (without ever converting the continuous analog wheel sensor signals to digital signals or before such conversion).

Peak detection

[0094] For each detected wheel pulse, the precise location of the peak (or valley) corresponds to the time of arrival of each individual wheel centre. This time of arrival can be estimated using a peak detection method. Whilst not necessarily shown in the earlier figures, when the analog signal is digitised, the digitised wheel sensor signal will be a series of discrete data points. The true peak of the wheel pulses is usually not captured in the discrete data points. Thus, the peak detection method seeks to identify the peak from the discrete data points which collectively serve as peak indicia without individually identifying the peak. Any peak detection method that is suitable to detect the peak from the wheel sensor output can be employed to estimate the time of arrival of each individual wheel centre.

[0095] For example, Figure 8 illustrates a peak detection method using quadratic interpolation, which can be used to locate the peaks (valleys) 88 of the wheel sensor outputs accurately by fitting a quadratic curve to the digital points 85 which represent the filtered digital outputs, and which generally do not coincide with the exact peak. A person skilled in the art will appreciate that other data fitting algorithms may be employed to identify the peaks (valleys) from the data. One suitable method for peak finding includes a LabVIEW built in function - Peak Detect VI - which is used to locate the peaks. This peak-finding algorithm consists of fitting a parabola to successive groups of points and interpolating the local peaks (or valleys).

[0096] In alternative embodiments, suitable analog peak detectors can be used to identify the peaks in the continuous analog wheel sensor signals.

Speed and Angle of Attack estimation

[0097] The speed can be calculated from the time of arrival of a wheel passing the second and third wheel sensors 12, 13 on the second rail 8 and the known predetermined distance between these two sensors 12, 13. Thus, the speed calculation can be expressed as:

¾1-S2

V = tsi t _s 2 where, V = speed of rail vehicle

D _S1-S 2 = known distance between second and third sensors t _sl - ts2 = difference in time of arrival of second wheel at the second and third sensors

[0098] The angle of attack may be derived by comparing the time of arrival of wheels of a wheelset, i.e. the time of arrival estimated from the first wheel sensor 10 on the first rail 6 and the time of arrival from the second wheel sensor 12 on the second rail 8. With reference to Figure 9, it will be appreciated that the angle of attack can also be expressed as the angle between the radial line relative to railway track and the axle of the wheelset (i.e. this is the same angle as in Figure 1 but measured from a different perspective). The angle of attack calculation can therefore be expressed as:

L — V ( t _WSi twS2 ) where, l = distance deviation of the second (outer) wheel relative to the first (inner) wheel

D = the distance between first and second wheel flange tops t _sl - t _S 2 = difference in time of arrival of second and third wheels at the second and third sensors t _wsl - t _WS2 = difference in time of arrival of both inner and outer wheel Angle of attack offset (calibration)

[0099] In practice, the pair of wheel sensors used for measuring the angle of attack will not be perfectly aligned along an imaginary line that is perpendicular to the longitudinal axis of the rail after installation. Any deviation from the ideal alignment will cause an error or “offset” in the estimated angle of attack values. However, this deviation can be suitably compensated for during this stage of the process. If the pair of wheel sensors are installed on a curved section of track, a known attribute of trail wheelsets of rail vehicles installed with rigid bogies is that they will generally align themselves on the track (to the curve centre) and achieve nearly Ό’ angle of attack. Thus, it is possible to utilise the measured angle of attack from trailing wheelsets of rigid bogies to offset the measured angles of attack values. For example, an average of the angles of attack determined for trailing wheelsets can be used for this calibration. This could involve firstly assessing whether the angle of attack values of the trailing wheelsets are for good steering axles, and using only those associated angle of attack values for the averaging. If the pair of wheel sensors are installed on a tangent section of track, it is possible to utilise a combination of measured angle of attack values for both trailing and leading wheelsets of rigid bogies to offset the measured angles of attack values (e.g. taking a suitable average). Again, this could involve firstly assessing whether the angle of attack values of the wheelsets are for good steering axles, and using only those associated angle of attack values for the averaging. This could be done on a wagon by wagon basis, a rail vehicle by rail vehicle basis, or by taking an average across multiple rail vehicles in a period being analysed in the digital signals.

[0100] It will be readily appreciated that a similar approach can also be adopted when working directly from the continuous analog wheel sensor signals.

Verification and accuracy of system through testing

[0101] Through testing of system 100, the inventors have been able to verify its overall accuracy. The four main variables influencing the angle of attack measurement are: (1) sensor spacing, (2) variation of wheel flange distance, (3) the time of arrival of the wheel centres, and (4) wheel sensor misalignment during installation.

[0102] Using the Frauscher RSR110 wheel sensor (having two built-in systems) during testing, the sensor spacing as previously mentioned is 131 mm, with tolerances of ±0.1 mm. The percentage error introduced by the sensor spacing tolerances is calculated as 0.08% (±0.1 mm/ 131 mm), which is used for analysing the relative error of the speed measurement.

[0103] For the assembly of all wheelsets, the wheelset back to back dimension is in the range 1357mm to 1360mm according to ASA standard RSU200 Series Section 6, Wheel and axle assembly- RSU 230. The distance between the back of a wheel and the tip of the wheel flange is 15 mm for both new WPR 2000 and ANZR1 wheel profiles according to ASA standard T HR RS 00870 ST RSU. As illustrated in Figure 11 , the average wheel flange distance D is calculated as:

(1357 + 1360)

D = + 15 + 15 = 1388.5 mm 2

The variation of the wheel flange distance, introduces a relative error 0.1 % (±1.5 mm/ 1388.5 mm), which is used for analysing the relative error of the angle of attack measurement.

[0104] Based on the type of wheel sensors used (Frauscher RSR110 wheel sensors) and the signal processing earlier described, the accuracy in measuring the time of arrival of wheel centres is within the data sampling interval. For example, for a given data sampling rate of 25600 Hz, the interval between two consecutive data points is 0.039 ms (1/25600 s). The accuracy of the time of arrival of wheel centres is ±0.0195 ms for a 25600 Hz sampling rate. It will be appreciated that more accurate estimation of the time of arrival of the wheel centres can be achieved using a higher data sampling rate. A 25600 Hz sampling rate was selected in testing as it provided the optimal balance of computing power, data size requirements and cost, for a non-permanent pilot system.

[0105] Wheel sensor misalignment during installation is accounted for with an angle of attack offset as earlier described.

[0106] The accuracy of the speed calculation can be estimated using the sensor spacing, and the time of arrival measured from first and second wheel sensor sub system (i.e. two built-in systems of the Frauscher RSR110 wheel sensor). Based on the law of error propagation, the accuracy of the speed measurement can be determined by the following formula: where o _speed is the accuracy of speed measurement for a given speed;

^s _pacing is the accuracy of sensor spacing (±0.1 mm); the sensor spacing is 131 mm; a _ws is the accuracy of time of arrival measurement, which is ±0.0195 ms for a 25600 Hz sampling rate;

Time is the time duration for a wheel to travel 131 mm at a given speed.

[0107] Table 1 presents the calculated accuracies for speeds ranging from 5 km/h to 80km/h, for a sampling rate of 25600 Hz. It shows that the measured speed is within ± 0.4km/h for speeds less than 80km/h.

Speed Spacing s Spacing Time otime s speed Speed relative

(km/h) (mm) (mm) (ms) (ms) (km/h) error

5 131 0.1 94.320 0.0195 0.0 0.1%

10 131 0.1 47.160 0.0195 0.0 0.1%

15 131 0.1 31.440 0.0195 0.0 0.1%

20 131 0.1 23.580 0.0195 0.0 0.1%

25 131 0.1 18.864 0.0195 0.0 0.2%

30 131 0.1 15.720 0.0195 0.1 0.2%

35 131 0.1 13.474 0.0195 0.1 0.2%

40 131 0.1 11.790 0.0195 0.1 0.2%

45 131 0.1 10.480 0.0195 0.1 0.3%

50 131 0.1 9.432 0.0195 0.2 0.3%

55 131 0.1 8.575 0.0195 0.2 0.3%

60 131 0.1 7.860 0.0195 0.2 0.4%

65 131 0.1 7.255 0.0195 0.3 0.4%

70 131 0.1 6.737 0.0195 0.3 0.4%

75 131 0.1 6.288 0.0195 0.3 0.4%

80 131 0.1 5.895 0.0195 0.4 0.5%

Table 1 - Accuracies of speed measurement

[0108] To determine the accuracy of the angle of attack measurement, the angle of the attack calculation can be expressed as follows: [0109] Considering the time difference between the two opposite wheel sensors,

(t _ws 1 t _WS2 ) or (t _sl - t _S2 ), are significantly less than the other items shown on the the right side of the above equation, the accuracy analysis of the AoA measurement can be simplified as follows: where s _AoA is the accuracy of AoA measurement; a _ws is the accuracy of time of arrival measurement, which is ±0.0195 ms for a 25600 Hz sampling rate;

D _S1 -S2 is the sensor spacing, 131 mm;

D is the flange distance 1388.5mm;

Time is the time duration for a wheel to travel 131 mm at a given speed; s _AoA offs t is the accuracy of the AoA-Offset.

[0110] Table 2 lists the estimated angle of attack accuracies for various speeds. The angle of attack measurement accuracy decreases slightly with increasing speed, and is generally within ±0.4 mrad for speed up to 80 km/h.

Speed Spacing Flange D Time otime

(km/h) (mm) (mm) (ms) (ms) a AoA (mrad)

5 131 1388.5 94.320 0.0195 0.0

10 131 1388.5 47.160 0.0195 0.1

15 131 1388.5 31.440 0.0195 0.1

20 131 1388.5 23.580 0.0195 0.1

25 131 1388.5 18.864 0.0195 0.1

30 131 1388.5 15.720 0.0195 0.2

35 131 1388.5 13.474 0.0195 0.2

40 131 1388.5 11.790 0.0195 0.2

45 131 1388.5 10.480 0.0195 0.2

50 131 1388.5 9.432 0.0195 0.3

55 131 1388.5 8.575 0.0195 0.3

60 131 1388.5 7.860 0.0195 0.3

65 131 1388.5 7.255 0.0195 0.4

70 131 1388.5 6.737 0.0195 0.4 75 131 1388.5 6.288 0.0195 0.4

80 131 1388.5 5.895 0.0195 0.4

Table 2 - Accuracies of AoA measurement

[0111] When wheelsets roll over two closely spaced pairs of wheel sensors (such as the two built-in systems of the Frauscher RSR110 wheel sensor), it is expected that the angle of attack of the majority of the wheelsets will experience steady curving, and therefore no or little change in angle of attack. However, in a very small number of instances, axles may experience unsteady curving, and the angle of attack may vary dynamically. By examining and comparing angle of attack measurements from two independent built-in systems which are closely spaced (131 mm), the accuracy and precision of the angle of attack measurements can be verified. For a given spacing of 131 mm between two pairs of wheel sensors, the theoretical difference in AoA can be calculated using the formula:

AoAl — AoA2 = Wheel Senor Spacing /Curve radius = 131 mm / 300 m = OA mrad

[0112] Examples of angle of attack differences measured from both passenger and freight trains are shown in Figures 12 and 13 respectively. It can be seen in both examples that the measured angle of attack differences are generally within a small range of ±0.5 mrad, and correlate well with the theoretical angle of attack difference of 0.4 mrad. There are a few outliers with an angle of attack difference greater than ±1 mrad from the freight train passby as shown in Figure 13 - axle numbers 41 and 125. Upon review of the data, it was found that the large angle of attack differences were due to increased speed variation between the inner and outer wheels of the particular wheelset.

[0113] The distribution of the angle of attack differences were analysed using a large dataset of more than 1 million axle passes, and over a period of 6 months. The dataset includes angle of attack measurements from all passenger and freight trains for that 6 month period. The results are presented in the histogram of Figure 14, which shows a typical bell-shape distribution having an average angle of attack difference of 0.2 mrad and a small standard deviation of 0.3 mrad. This result confirms that the angle of attack measured from two separate wheel sensor pairs agree well with each other, and are within a tight range 0.2 ± 0.3 mrad, or -0.1 to 0.5, with the theoretical value 0.4 mrad. Figure 14 confirms that angle of attack measurment accuracy of within ± 0.4 mrad can be achieved by system 100.

[0114] As an extension of system 100, reference is now made to Figure 15 and wheelset hunting detection system 200. Angles of attack can be measured at multiple locations along a railway track to detect the hunting condition of a wheelset, bogie or vehicle. As previously mentioned, hunting is a condition in which a wheelset, bogie or rail vehicle swings from side to side between rails while the train operates at speed on a tangent section of track. Hunting may be caused by worn components, wear or defects in the track or rails or a variety of other reasons. Hunting results in rapid wear of vehicle components, rails and other railway components, and has the potential to cause damage to cargo and perhaps eventually lead to derailment. If the angle of attack of the wheelset changes as it passes over each first sensor 10 or second sensor 12, this may indicate a hunting wheelset condition is present. In addition, the greater the change in angle of attack of a wheelset between sensors along a rail, the greater the likely severity of hunting. In one example, if the change in angle of attack of a wheelset between various points along the track is considered to be outside of a predetermined threshold or level, a hunting wheelset condition may be indicated. This indication may be in the form of an alarm (visual and/or audible) or the hunting condition may be suitably logged for later inspection. In one example, an email, SMS or other form of warning message is sent when a wheelset/bogie/wagon is detected as hunting.

[0115] Multiple pairs of wheel sensors 10, 12 are positioned at predetermined distances along the rail for detecting hunting. Each pair of wheel sensors 10, 12 are aligned perpendicularly to the longitudinal axis of a rail for measuring angle of attack in much the same way as mentioned earlier. The number of pairs and the distance between each adjacent pairs can be determined at least in part by the distance between wheelsets of bogies, and the distance between bogies of rail vehicles. However, any suitable number of pairs of wheel sensors 10, 12 and/or suitable distances between the pairs of sensors can be used. The hunting condition of a wheelset, bogie or vehicle can be determined by comparing the measured angles of attack of each individual wheelset at various points along the track. [0116] It will be appreciated from Figure 15, that system 200 is implemented using the same general layout as system 100, i.e. by using a single data acquisition unit 16 and computer processing system 500 for storing and processing the signals to measure the angles of attack. However, the system could include multiple data acquisition units and computer processing systems, e.g. one for each sensor pair. The same processing steps mentioned previously would also apply in system 200. It is noted that for hunting detection, a higher sampling rate may be used when digitising the analog signals due to the higher speed of travel of the rail vehicle on a tangent section of track.

[0117] Reference is now made to Figure 16, which provides the signal processing steps of Figure 5, except with another series of signal processing steps that can be undertaken in combination with or independently of the earlier steps in order to determine a wheel and/or rail vibration condition, e.g. corrugation, squealing, etc. These other steps generally involve the following taking place with the digital signal: frequency analysis 58 and wheel and/or rail condition detection 59. The monitoring and detecting of a wheel/rail vibration condition is important in mitigating wheel/rail induced vibration and noise. A further or more accurate determination of wheel and/or rail condition 60 (as will be described below) can be provided by also incorporating the determined angle of attack in the processing.

[0118] As shown in Figure 16, before the offset wheel sensor signals are filtered, the unfiltered signals can be analysed using suitable frequency analysis methods (e.g. Fast Fourier Transform analysis) to identify the frequency components of the digitised wheel sensor signals. The frequency components of the wheel sensor signals can then be used to determine the vibration condition of the wheel and/or rail.

[0119] With reference to Figure 17(a), an example is provided of frequency analysis being performed on a wheel sensor signal 132 that has been attained under conditions with substantially no vibration. Under such conditions, a very clean wheel sensor signal 132 is shown in the time domain, and in the frequency domain the magnitudes of all frequency components are very small.

[0120] When there is vibration present in the wheel sensor signals, a sharp rise of the frequency component at the corresponding frequency ranges will be present. For example, as shown in Figure 17(b), wheel sensor signal 134 has corrugation level frequencies (typically in the range of about 50 Hz to about 1000Hz). In Figure 17(c), wheel sensor signal 136 has squeal level frequencies (typically in the range of about 1000 Hz to about 10,000Hz). By comparing the relative magnitudes of the frequency components of the wheel sensor signal, a determination can be made as to the wheel/rail vibration condition.

[0121] In one example, a comparison of the relative magnitudes of the frequency components for each individual wheel pulse can be undertaken to determine (or provide an indication of) its vibration condition, and the type of vibration. For example, it can be determined whether the wheel/rail condition is a wheel squeal condition (higher frequency range as indicated above) or a corrugation condition (lower frequency range).

[0122] This information can be supplemented by comparing the vibration condition of two wheels of the same wheelset. This can enable determination of the source of vibration, i.e. whether it is from the inner wheel or the outer wheel (and which rail). In some instances, it may be determined that the vibration condition exhibited by, for example, the inner wheel is a result of one or more of: operative condition of the inner wheel, operative condition of the inner rail, operative condition of the outer wheel (transferring over to the vibration condition of the inner wheel), or operative condition of the outer rail.

[0123] Further information can also be provided by comparing the vibration condition of consecutive wheels, with vibration of similar frequency components being detected to determine the source of vibration, e.g. the wheel(s) with the highest vibration may be considered the main cause of the vibration condition of a given bogie, wagon or rail vehicle. In other examples, if vibration conditions appear to be consistent across multiple successive wheelsets, this may indicate that it is the condition of one or both rails that are the primary cause of the detected vibration.

[0124] It will be further appreciated that the determined wheel/rail condition can be used in conjunction with the determined angle of attack of a given wheelset to further inform the wheel/rail conditions as well as their source. A large angle of attack is regarded as necessary for corrugation or wheel squeal to occur. However, not all wheels with large angles of attack excite corrugation or squeal. The wheel/rail vibration condition detection enables identification of the source of vibration or noise, and does not necessarily mean that wheel/rail defects are present. For example, wheel/rail vibration can occur at the corrugation frequency without obvious rail corrugation, and wheel squeal can occur without obvious wheel defects.

[0125] The monitoring and detection of wheel/rail vibration in conjunction with angle of attack can help determine the source of vibration to direct noise/vibration mitigation. For example, the determined wheel/rail condition can help inform whether an unusually high angle of attack is attributable to corrugation (indicating rail defects), squeal (indication wheel defects), or possibly point to some other issue. For example, information can be collated for different wheelsets, different bogies/wagons/rail vehicles, and this information can provide a more complete picture of the wheel/rail conditions that are monitored in the system.

[0126] As shown above, in one example the present invention has been shown to achieve accuracy of angle of attack determination within ± 0.4 mrad for train speeds up to 80km/h. Higher accuracy can potentially be achieved in some cases with higher sampling rate of the data. Further, certain embodiments of the present invention may be able to capture and calculate angle of attack for almost all passing axles, with minimal loss of data using the techniques earlier described.

[0127] It will be understood that the invention disclosed and defined in this specification extends to all alternative combinations of two or more of the individual features mentioned or evident from the text or drawings. All of these different combinations constitute various alternative aspects of the invention.