GB2549539A

GB2549539A - Method and apparatus for processing wheel phase angle sensor signals in a tyre pressure monitoring device

Info

Publication number: GB2549539A
Application number: GB1607080.7A
Authority: GB
Inventors: Mckellar Robert; Molyneaux Nevin
Original assignee: Schrader Electronics Ltd
Current assignee: Schrader Electronics Ltd
Priority date: 2016-04-22
Filing date: 2016-04-22
Publication date: 2017-10-25

Abstract

A method of processing an output of a wheel phase angle sensor of a wheel monitoring unit is disclosed, wherein: the sensor output is time sampled, the time samples being pre-processed and transformed into the frequency domain using a frequency transform such as Fast Fourier Transforms (FFT). The frequency transform coefficient corresponding to the fundamental frequency of the sensor output is identified and its phase is calculated. The calculated phase is taken as the phase of the sensor output, which may then be used in wheel auto-location on a vehicle. This method can be used in Wireless Auto Location (WAL) or Phase Auto Location (PAL) systems. The phase difference between two sets of time samples may be calculated to increase accuracy of the frequency of the sensor output signal. An All phase Fast Fourier Transform (ApFFT) may be performed on the data sample. A triangular window function may be applied to the sample data. Also claimed is a wheel monitoring unit comprising means to carry out the method.

Description

Method and Apparatus for Processing Wheel Phase Angle Sensor Signals in a Tyre Pressure

Monitoring Device

Field of the Invention

The present invention relates to wheel monitoring units, particularly for Tyre Pressure Monitoring Systems (TPMS). The invention relates particularly, but not exclusively, to wheel mountable tyre pressure monitoring devices for use in a TPMS that performs wheel auto-location.

Background to the Invention

Tire pressure monitoring systems generally include a wheel monitoring unit in the form of a tire pressure monitoring (TPM) device (commonly referred to as a TPM sensor) in or at each wheel of a vehicle, and a central controller which receives tire pressure information from each TPM sensor, for reporting to the driver of the vehicle. Auto-location involves identification of each TPM sensor and determination of its position on the vehicle, automatically and without human intervention. Autolocation may be done initially upon installation and subsequently in the event of tire rotation or replacement. Performing auto-location typically involves determining the identity, e.g. by serial number, of a TPM sensor in each of the wheels in the car. Knowing the identity of the TPM sensor in each wheel allows a pressure by position display to be implemented and shown to the driver, while in vehicles with different placard tire pressures for front and rear axles, it is desirable to know TPM sensor identities and positions in order to check pressure against a correct threshold for an applicable axle.

International patent application WO 2011/038033 and United States patent application US 2013/0079977 each disclose a TPMS system in which the central controller uses wheel phase information transmitted to it by the TPM sensors to perform auto-location of the sensors. These systems are of a type that may be referred to as Phase Auto-Location (PAL) systems. US patent US 7010968 discloses a TPMS system that performs a different type of auto-location known as Wireless Auto-location (WAL). During WALtype auto-location, the phase relationship between two shock sensors is analysed, allowing the TPMS sensor to establish whether it is rotating clockwise or anticlockwise. During PAL type auto-location, the phase of a single shock sensor output is correlated with a vehicle’s ABS data, enabling auto-location.

In a TPMS that performs auto-location, the wheel mounted unit includes a wheel phase angle sensor, for example a shock sensor, that produces a signal that is indicative of the angular position of the wheel. As the sensor rotates with the wheel, it tends to produce a sinusoidal output signal. In practice, the output signal produced by such sensors rarely resembles a perfect sinusoid. Imperfect road surfaces tend to result in what appears to be random noise being superimposed on the signal.

Additionally, it has been observed that the output signal includes one or more additional frequency components as well as a fundamental frequency component.

The presence of noise and other unwanted signal components makes it difficult to extract the following information from the shock sensor output: a) the phase difference between two shock sensor fundamental sinusoids (WAL) b) the frequency and phase of the shock sensor fundamental sinusoid (PAL)

Unwanted signal components such as noise can be filtered using a tunable analogue filter. The filtered signal may then be converted to digital signals which can be analysed by a suitably programmed processor. However, such filters introduce a phase delay which varies with vehicle speed. Also, step inputs to the shock sensor as a result of large tyre impacts with the road surface (e.g. pot hole) can result in filter saturation. The TPMS sensor must then wait until the filter has recovered before any more shock sensor information can be obtained. Furthermore some of the unwanted signal components are often close in frequency to the frequency of the fundamental signal component meaning that the noise can pass through the filter with little attenuation.

It would be desirable therefore to provide an improved method and apparatus for processing wheel phase angle sensor outputs in a TPMS sensor.

Summary of the Invention A first aspect of the invention provides a method of processing an output of a wheel phase angle sensor of a wheel monitoring unit mounted on a wheel to determine a phase value for said wheel phase angle sensor, the method comprising: sampling, when said wheel is rolling, said output to produce time sampled output values; storing a plurality of said time sampled output values; performing a frequency transform on said stored output values, or a plurality of values derived therefrom, to produce a plurality of frequency transform values corresponding to respective frequency components of said wheel phase angle sensor output; identifying one of said frequency transform values as corresponding to a fundamental frequency component of said wheel phase angle sensor output; and determining said phase value from the identified frequency transform value.

Preferably, said frequency transform values are represented as a respective complex number, said determining said phase value comprising calculating the phase of the complex number corresponding to said identified frequency transform value.

Preferably said arranging involves providing a plurality of series of data values in which said centre sample value is the first in each series, said FFT being applied to each series to produce a respective set of frequency coefficient values, said frequency coefficient values being averaged to produce said frequency transform values. Typically N series of data values are provided, each comprising N data values including the centre sample value, and said FFT is an N-point FFT.

Preferably, each series of data values contains a respective set of said time sampled output values, the sets being time shifted with respect to each other and overlapping such that each set includes the centre sample value, and each other time sampled output value is included in at least one set.

In preferred embodiments, the method further includes determining a value for the fundamental frequency of said wheel phase angle sensor output by: determining an estimate of said fundamental frequency value from said identified frequency transform value; determining the respective phase value for first and second sets of time sampled output values, the sets being spaced apart in time by at least one sample period; calculating a frequency correction value from the difference between the respective phase values; and adjusting said estimate of said fundamental frequency value using said frequency correction value to calculate said fundamental frequency value.

Preferably said first and second sets of time sampled output values are spaced apart by N samples, and wherein calculating said frequency correction value involves multiplying the phase difference by Ρ$/(2πΝ), where Fs is the sampling frequency.

The preferred method further includes calculating a phase per sample value by multiplying the calculated fundamental frequency value by π/Fs, and determining a phase value for any sample value of the first or second time sampled output values using said phase per sample value and linear extrapolation. A second aspect of the invention provides a method of performing wheel auto-location, the method including the method of the first aspect of the invention. The preferred method includes processing a respective output of first and second wheel phase angle sensors of a wheel monitoring unit using the method of the first aspect of the invention. The preferred method also includes processing an output of a single wheel phase angle sensor of a wheel monitoring unit to calculate the phase value and fundamental frequency value using the method of the first aspect of the invention. A third aspect of the invention provides a wheel monitoring unit comprising: at least one wheel phase angle sensor, the method comprising: means for sampling, when said wheel is rolling, said output to produce time sampled output values; means for storing a plurality of said time sampled output values; means for performing a frequency transform on said stored output values, or an a plurality of values derived therefrom, to produce a plurality of frequency transform values corresponding to respective frequency components of said wheel phase angle sensor output; means for identifying one of said frequency transform values as corresponding to a fundamental frequency component of said wheel phase angle sensor output; and means for determining said phase value from the identified frequency transform value.

Preferably, the wheel monitoring unit comprises means for performing the steps of any one of the methods of the first or second aspects of the invention. A fourth aspect of the invention provides a tyre pressure monitoring system comprising at least one wheel monitoring unit of the third aspect of the invention, preferably a respective wheel monitoring unit for each wheel of a vehicle.

Preferred embodiments of the invention allow the frequency and phase of the fundamental signal component (sinusoid) to be extracted from the output of a shock sensor, or other wheel phase angle sensor, despite the presence of noise.

Further advantageous aspects of the invention will be apparent to those ordinarily skilled in the art upon review of the following description of a specific embodiment and with reference to the accompanying drawings.

Brief Description of the Drawings

An embodiment of the invention is now described by way of example and with reference to the accompanying drawings in which:

Figure 1 is a schematic view of a tyre pressure monitoring system incorporated into a vehicle;

Figure 2 is a schematic view of a wheel monitoring unit suitable for use with the system of Figure 1 and embodying the invention;

Figure 3 is a schematic view of part of the wheel monitoring unit, showing those components used in processing the output of a wheel phase angle sensor;

Figure 4 is a flowchart illustrating two embodiments of an all-phase FFT pre-processing method, the method embodying one aspect of the invention and being suitable for use with other aspects of the invention;

Figures 5A, 5B, 5C, 5D and 5E, show for an N=4 example, alternative steps for sorting sample data in preferred embodiments of the all phase FFT pre-processing method;

Figure 6A is a schematic diagram of an all-phase FFT architecture;

Figure 6B shows preferred coefficient values for use during a windowing operation that is included in the preferred pre-processing method;

Figure 6C shows a data vector formed by the preferred windowing operation;

Figure 7 shows a representation of an output frequency component of a Fourier transform as a complex number; and

Figures 8A and 8B illustrate an example of wheel phase angle sensor output in the frequency domain.

Detailed Description of the Drawings FIG. 1 Illustrates a wheel monitoring system in the particular form of a tire pressure monitoring system 100. The system 100 is shown incorporated in a standard vehicle 1 having four wheels, namely a left front wheel (LF), a right front wheel (RF), a left rear wheel (LR) and a right rear wheel (RR), although it may alternatively be used with a vehicle having more or fewer wheels. The system 100 includes wheel units 101, 102, 103 and 104 each being associated with a respective wheel of the vehicle 1.

In typical embodiments, the system 100 includes antilock brake system (ABS) sensors 201,202, 203 and 204, usually one for each wheel. In this embodiment, ABS sensors 201,202, 203, 204 are each associated with a respective wheel of the vehicle 1. In other embodiments, ABS sensors may not be associated with all of the vehicle’s wheels.

The system 100 includes a central controller in the form of an Electronic Control Unit (ECU) 300, and an associated wireless receiver 400 (which may comprise a transceiver) for receiving transmissions from the wheel units 101, 102, 103, 104. The ECU 300 Is coupled to the ABS sensors 201,202, 203, 204 via a communication bus such as a Controller Area Network (CAN) bus and receives ABS data from the ABS sensors 201 202, 203, 204. The ECU 300 includes a processor 302 and data storage 304. The ECU 300 may be implemented by any suitable means, for example a microprocessor, microcontroller, an Application Specific Integrated Circuit (ASIC), or other suitable data processing device programmed to perform the functions described herein.

In use, the ECU 300 receives data from the wheel units 101,102, 103 and 104 via the receiver 400. For example, the wheel units 101, 102, 103 and 104 may be configured to transmit radio frequency or other wireless communication signais conveying data and other information to the ECU 300. The respective wheei units inciude a suitabie radio (or other wireiess) transmission circuit and the ECU 300 inciudes a suitabie radio (or other wireiess) reception circuit for wireiess communication. The system may be equipped to provide two-way communication between the ECU and the wheei units in which case each may inciude a suitabie transceiver rather than just a receiver or transmitter.

The system 100 is configured to perform auto-location of the wheei units 101, 102, 103, 104 using data transmitted from the units 101, 102, 103,104 to the ECU 300. in preferred embodiments, the transmitted data inciudes wheei phase angie information and so the system 100 may be referred to as a Phase Auto-Location (PAL) system, for exampie of the type disciosed in WO 2011/038033 or US 2013/0079977. As such, the ECU 300 is configured to correiate the data received from the wheei units 101, 102, 103 and 104 with ABS data received from the ABS sensors in order to perform auto-iocation, for exampie in the manner described in WO 2011/038033 or US 2013/0079977. Alternativeiy, the system 100 may be configured to perform Wireiess Auto-Location (WAL), for exampie of the type disciosed in US 7010968. in aiternative embodiments, the system may include one or more sensors, devices, sub-systems, or mechanisms that provide wheel phase and/or speed data for use in addition to, or instead of data suppiied by an anti-lock brake system when performing auto-iocation. The system 100 is typicaiiy operabie in more than one mode, including for example a first mode in which it performs a task such as tire pressure monitoring, and a second mode in which it performs auto-iocation. in generai, the controiier222 receives data generated by the various sensors 206, 208, 212 (usuaiiy after some signai processing has been performed) and processes the data in order to provide reievant output signais for transmitting to the ECU 300. The controller 222 may process the data itself, for example using computer software in cases where the controller 222 comprises, say a microcontroller or microprocessor, and/or using electronic hardware, such as one or more digital signal processor(s), which may be considered to be part of the controller 222 or otherwise as resources available to the controller 222, depending on the implementation.

Referring to FIG. 2, the structure of a typical wheel unit 101 is described in more detail. The wheel units 102-104 may incorporate the same structure as that of the wheel unit 101. The illustrated wheel unit 101 includes a controller 222, a battery 204 (or other electrical power source), a transponder coil 206, a sensor interface 207, a pressure sensor 208, at least one wheel phase angle sensor 212, a transmitter 214 and an antenna 216. In other embodiments, the wheel unit 101 may have a different structure from the structure illustrated in FIG. 2, and some of the illustrated components may be omitted depending on the application. For example, in the present embodiment it is assumed that the system 100 is operable as a tyre pressure monitoring system and so the wheel unit 101 includes a pressure sensor 208 for monitoring the pneumatic pressure in the wheel’s tyre. In other embodiments, the system may be configured to monitor one or more other characteristics of the wheel, for example temperature, in which case it may include one or more other sensors, for example a temperature sensor, instead of or as well as the pressure sensor 208. The controller 222 may be implemented by any suitable means, for example a microprocessor, microcontroller, an Application Specific Integrated Circuit (ASIC), and/or other suitable data processing device and/or circuitry programmed to perform the functions described herein, and may include or have access to suitable data storage as required.

The controller 222 is coupled to the sensor Interface 207. The sensor interface 207 is coupled to the wheel phase angle sensor 212. The wheel phase angle sensor 212 produces an output signal that Is indicative of the angular position of the wheel as the wheel rotates, the output signal being provided to the sensor interface 207. It may be said that the wheel phase angle sensor 212 provides wheel phase angle measurements, i.e. data Indicating the angular position of the wheel as the wheel rotates, to the sensor interface 207. Alternatively, or additionally, the wheel phase angle sensor 212 provides other value(s) or information indicative of wheel phase angle. The sensor interface 207 receives the output of the wheel phase angle sensor 212 in the form of an electrical signal. The sensor interface 207 receives the electrical signal and processes it, the processing typically involving amplification and optionally filtering. As such, the interface 207 may include one or more suitable filters and one or more suitable amplifiers (not shown in Figure 2).

The sensor interface 207 may send the processed signal to an analog to digital converter (ADC) (not shown in Figure 2) in order to convert the signal into a digital signal. Alternatively, the signal may be converted to digital before it is processed, and then processed digitially by a suitable digital signal processor. In either case, the ADC may be part of the sensor interface 207 or may be provided separately, e.g. as part of an ASIC, as is convenient. The controller 222 receives a digital form of the signal from the wheel phase angle sensor 212 for processing as is described in more detail below.

In the illustrated embodiment, the pressure sensor 208 detects the pneumatic air pressure of the tire with which the wheel unit 101 is associated. In alternative embodiments, the pressure sensor 208 may be supplemented with or replaced by a temperature sensor or other devices for detecting tire data. An indication of the tire pressure data is sent to the controller 222 via an analog-to-digital converter (not shown).

The battery 204 is a power source of the wheel unit 101. The transponder coil 206 detects external activation of the transponder by a signal applied by a remote exciter and may modulate a signal to communicate data to a remote detector from the wheel unit 101. The wheel unit 101 provides data including tire pressure from the pressure sensor 208 and the wheel phase angle information from the wheel phase angle sensor 212 through the transmitter 214 and the antenna 216 to the ECU 300.

Upon rotation of a wheel, the wheel phase angle sensor 212 operates to measure the wheel phase angle, i.e. provides a signal indicating the angular position of the wheel at any given time during rotation. Wheel phase angle measurements need not necessarily be made with respect to an absolute reference. A reference may be arbitrarily selected based on accuracy capability and ease of implementation. In other words, the phase measurements do not have to be measured from a top of wheel, or road striking point or other specific point. In some embodiments, the key piece of information required by the system 100 may be a phase difference, or a phase delta of the wheel. and therefore, the requirement may be that two different phase angles are measured relative to the same reference angle. Alternatively, or additionally, in other embodiments, the key piece of information may comprise a single angle measurement during a rotation of a wheel.

The wheel phase angle sensor 212 (and more generally the wheel mounted units) may be mounted in any suitable location on or in the wheel, for example on a rim of the wheel, e.g. coupled to the tyre’s valve, or on the tyre itself, typically on an internal surface of the tyre. In one embodiment, the wheel phase angle sensor 212 comprises a rotation sensor. For example, the rotation sensor may be a piezoelectric rotation sensor which measures a wheel phase angle based on gravitational force experienced by the sensor. Specifically, as the wheel rotates, the gravitational force causes a sensing element of the rotation sensor to experience different forces which results in a different output signal representing a wheel phase angle (or wheel angular position). In that way, the rotation sensor produces an output signal indicating a wheel phase angle overtime as the wheel rotates. The output signal of the rotation sensor may have different amplitude and/or different polarity depending on the wheel phase angle. For instance, the rotation sensor may produce an output signal having amplitude M at 0 degrees and having the amplitude -M at 180 degrees. Alternatively, or additionally, any conventional rotation sensor may be used as the wheel phase angle sensor 212.

Alternatively, the wheel phase angle sensor 212 comprises a shock sensor of the type that produces an electrical signal in response to changes in acceleration. The electrical signal is indicative of, typically proportional to, the experienced change in acceleration. Alternatively, the wheel phase angle sensor 212 may comprise an accelerometer or a micro-electromechanical systems (MEMS) sensor. In any case, in preferred embodiments, the wheel phase angle sensor 212 produces a sinusoidal output signal in response to rotation of the wheel on which it is mounted, or more particularly an output that includes a fundamental sinusoidal component. Alternatively, the wheel phase angle sensor may produce an output signal that may be represented sinusoidally, e.g. after suitable processing. For example, the wheel phase angle sensor may alternatively comprise a Hall Effect sensor or a road strike sensor.

Typically, the output signal of the sensor 212 comprises a sinusoid with a period equal to one revolution of the wheel. More particularly, one cycle of the fundamental sinusoidal component of the sensor 212 output corresponds to one revolution of the wheel. The magnitude of the output signal is a voltage proportional to the change In acceleration or acceleration experienced by the wheel phase angle sensor 212 as it rotates. The controller 222 may be configured to recognize a repeating pattern in the signal produced by the sensor 212 and so to identify respective rotations of the wheel.

Hence, via the sensor interface 207, the controller 222 receives signals representing wheel phase angle from the wheel phase angle sensor 212. The controller 222 determines wheel phase angle data from the received signal to enable auto-location to be performed by the ECU 300. The controller 222 typically stores the wheel phase angle data for transmission to the ECU 300 at an appropriate time. The controller 222 encodes and transmits data via the transmitter 214 and the antenna 216.

The data preferably includes wheel phase angle data, typically amongst other things such as tire pressure information and an identifier of the wheel unit 101.

Wheel phase angle data may be determined in respect of a measurement period which may correspond to one or more revolutions of the wheel. The controller 222 may not transmit every time the output signal has been received, or in respect of every wheel revolution. The nature of the wheel phase data gathered by the controller 222 depends on the auto-location scheme being implemented by the system and may for example be as described in WO 2011/038033, US 2013/0079977 or US7010968. For example, the controller 222 may determine a difference between two wheel phase angles measured at respective time instants in a wheel’s rotation. The controller 222 may determine a second time based on a predetermined known time delay from a first time. For instance, the controller 222 may consider the first time as a single measurement point of a wheel phase angle during the rotation of a wheel and the second time as a data transmission point for a communication to the ECU. The controller 222 may include a clock or time base, or other circuit or module for measuring time increments and operating at specified times or during specified time durations. The wheel phase angle data may include actual wheel phase angles measured at different times. In another embodiment, the wheel phase angle information may include a wheel phase angle measured at a transmission time, such as the second time, and a difference in wheel phase angle measured at two different times. Alternatively, the wheel phase angle data may include only the difference in wheel phase angles. In another embodiment, the wheel phase angle information may include no actual wheel phase angle. Instead, the wheel phase angle data includes a wheel phase angle indication.

Referring again to Figure 1, the ECU 300 receives the wireless communication from the wheel unit 201. The ECU 300 extracts wheel phase angle data from the communication along with any other relevant information e.g. the tire pressure and/or the wheel unit identifier. In some embodiments, the ECU 300 correlates the wheel phase angle data with the ABS data from the ABS sensors 201,202, 203, 204 for example as described in WO 2011/038033 or US 2013/0079977.

Figure 3 shows aspects of the wheel unit 101 in more detail, including only those components that are helpful for understanding the operation of the illustrated embodiment. In some embodiments, for example where PAL type auto-location is implemented, only one wheel phase angle sensor 212 is required. In other embodiments, more than one wheel phase angle sensors 212 are provided. For example where WAL auto-location is implemented, two wheel phase angle sensors 212 are provided. The various components of the wheel unit 101 that process the output of the sensor 212, including the sensor interface 207 and the relevant aspects of the controller 222 may be duplicated for each sensor 212 or shared amongst the sensors 212 as is convenient. The following description describes the operation of a preferred embodiment of the invention for a single wheel phase angle sensor 212. The same description applies for other wheel phase angle sensor(s) that may be provided depending on the embodiment. In the present embodiment, the wheel phase angle sensor 212 is assumed for illustrative purposes to be a shock sensor. However the same or similar description applies in embodiments where the wheel phase angle sensor 212 takes other forms, including those identified above, as would be apparent to a skilled person.

The sensor interface 207 in this example comprises an amplifier 224 and an anti-aliasing filter 226, each of which may comprise conventional analogue processing circuitry and components. The amplified and filtered analogue signal is provided to an ADC 228 for conversion into digital form. The digitised signal undergoes digital signal processing as indicated by component 230 in Figure 3. The digital signal processing is typically performed by and/or under the control of the controller 222 using one or more software routines and/or one or more digital signal processors (not shown) that may be part of the controller 222 or available to the controller 222 for digital signal processing. In typical embodiments, the amplifier 224, filter 226, ADC 228, controller 222 and any associated analogue or digital hardware used to perform the analogue and digital processing and storing of the sensor output signal are provided in an integrated circuit (IC), typically an ASIC.

The shock sensor output signal is relatively small (typically in the order of hundreds of microvolts) and so is amplified by the amplifier 224. In particular the amplifier 224 amplifies the output signal so that the signal range matches the input range of the ADC 228. Because it is desired to sample and store the shock sensor output signal, the filter 226, which is a low pass filter, is provided to prevent aliasing. The filter 226 should be constructed with the correct parameters so that all frequencies above the half sampling rate are attenuated below one count at the ADC, and preferably that the fundamental frequency of the shock sensors (which is typically between 1-40 Hz) is within the pass band of the filter.

The amplified shock sensor signals are sampled, typically at a fixed sampling frequency, over a predetermined period, typically between 1 and 5 seconds, for example a 1.5 second period or 3 second period, and the resulting sample values are stored. The signal sampling is typically performed under the control of the controller 222 using the sensor interface 207. The data sampling may be performed at any convenient rate, for example 100 Hz. The digitised sample values, i.e. as produced by the ADC 228 in this example, are stored in any convenient data storage device (not shown), typically under the control of controller 222. By way of example, 50 to 500 samples may be taken during the measurement period, depending on the length of the measurement period and the sampling rate.

The digital signal processing 230 is then carried out on the stored data sample values, usually offline (i.e. not in real time). In typical embodiments, all amplification, filtering, sampling, storage and digital processing is carried out within an Application Specific Integrated Circuit (ASIC), although external circuitry may alternatively be used to carry out any one or more of these functions.

The stored sample values are time domain sample values. The digital signal processing 230 involves transforming the time domain samples to the frequency domain. The transformation is preferably performed using a Fast Fourier Transform (FFT), most preferably an all-phase Fast Fourier Transform (ApFFT). Hence, in preferred embodiments, the shock sensor output signal is sampled at a fixed sampling frequency, over a pre-determined period of time, and digital signal processing (DSP) is used to transform the sample data from the time domain to the frequency domain. When the FFT is computed the resulting FFT values (also known as FFT coefficients) are stored in any convenient data storage device (not shown), typically under the control of controller 222. Each FFT value is generated as a complex number having a real part and an imaginary part. The FFT values are therefore stored in two arrays, one for the respective real part of the calculated complex numbers and one for the respective imaginary part. A respective set of FTT values is stored for each shock sensor channel required (e.g. one for PAL, two for WAL).

Once the sample data is in the frequency domain, an accurate determination of the phase and an estimate (for example to the nearest frequency bin) of the frequency of the fundamental frequency component of the shock sensor output signal are identified. This may be achieved using any one of various techniques including those described below.

An all phase FFT, unlike a standard FFT, cannot be calculated from a contiguous set of data samples. Instead, to calculate the ApFFT, the data is first pre-processed using a data sorting method, which produces a set of data to which a regular FFT can be applied. The phase accuracy limitations of a regular FFT are overcome by the pre-processing of the sorting method. In particular, by storing an input signal of length 2N-1 samples, it is possible to determine the phase of the centre sample in the series using an ApFFT.

The ApFFT can be performed by arranging the input signal of 2N-1 data samples (which in this case are samples of the shock sensor output) into a set of N different data vectors, each of which contains a slice of the input signal comprising a series of N sequential data samples, wherein one of the data samples in each data vector is the “centre” sample of the input signal. Normally the centre sample is (exactly) at the centre of the time series of sample values, i.e. the Nth sample of 2N-1 samples. However, this is not essential and more generally there should be enough data samples on either side of the selected centre sample to allow the data pre-processing (which is described in more detail below) to be performed. The N data vectors each cover a different slice of the original input signal, so that all 2N-1 data samples are captured. As each data vector contains N samples and must contain the centre sample, the data vectors overlap in time.

The data vectors are generated so that each contains the centre data sample. As an FFT measures phase at the first data sample in each vector, it is necessary to cyclically rotate any data vector which does not have the centre sample as the first sample so that the centre sample comes first. The purpose of sorting the data is to achieve zero phase error at the centre sample using the time/phase shift property of Fourier transforms.

Once a set of time shifted data vectors has been created and arranged so that the centre sample of the original time series appears first, an FFT being performed on each of the data vectors, the average value of the sum of the FFTs being equivalent to an all phase FFT. Alternatively, due to the linear properties of Fourier transforms, to save computational power instead of performing N separate FFTs, the data vectors can be averaged prior to the FFT calculation, resulting in a single average data vector on which one FFT is performed to find the phase of the centre sample of the original time series.

This is illustrated in the flowchart of Figure 4. Starting with 2N-1 samples at step 400, a set of N overlapping data vectors can be generated each of length N and including the centre sample (401). The data vectors are then rearranged, i.e. time/phase shifted, so the centre sample appears first (402). Subsequently, there are two mathematically equivalent methods of achieving an All phase FFT: either the N data vectors can each be transformed (403) and the results averaged (404) or the average of the data vectors can be taken prior to any transformation (405) and a single transform can be performed on the averaged data vector (406). In the preferred embodiment of the invention a value of N=64 is used as suitable choice for the TPMS application having the required accuracy level while not being overly resource intensive to perform. However it would be apparent to those skilled in the art that other values of N could be used, although in preferred embodiments N must be at least 2.

As an example, if N=4 then the input signal is 7 samples in length, with the centre sample the 4^^ sample in the series. A set of data vectors can be generated in a 4x4 matrix. Each data vector comprises a 4 sample section of the input signal. It will understood that the value N=4 is used for illustration purposes and that N may take any other suitable value.

Figure 5A shows, by way of example, a sample input x[n] of length 7 (2N-1, where N=4) samples, x[0] to x[6]. N=4 data vectors Xo[n] to Xs[n] are formed from respective segments of the input, each segment/vector comprising N=4 of the input data samples. Figure 5B shows the data vectors as 501, 502, 503 and 504; these data vectors comprise a respective set of the input data samples such that each vector contains a different portion of the input x[n]. There is significant overlap between the contents of the data vectors Xo[n] to Xsin], each vector including the centre data sample x[3].

Figure 5C shows a NxN matrix comprising the N=4 data vectors, each comprising N=4 data samples, including the centre sample. However it is necessary to move the centre sample so that it appears first in each vector Xo[n] to Xsin]. Due to the cyclic nature of Fourier transforms it is possible to rotate the vectors, i.e. to re-order their respective data samples.

Figure 5D shows the data of the matrix of Figure 5C re-arranged so that the centre sample x[3] is the first sample in each data vector. This re-arrangement creates a set of N=4 data vectors x®o[n] to x®3[n], on each of which an N=4 point FFT can be performed. Alternatively the data vectors x^o[n] to x®3[n] can be averaged creating one averaged data vector for which only one FFT need be performed.

As shown in Figure 5E a single averaged data vector XAp[n] can be constructed by averaging the set of re-arranged data vectors. In the example illustrated by Figure 5E, the averaging is performed by calculating an average sample value for each sample position in the averaged data vector by averaging the respective samples that are correspondingly positioned in each data vector x%[n] to x%[n].

Generation of XAp[n] may be performed using other methods. For example, in preferred embodiments the sample pre-processing involves using a set of windowing coefficients to apply a window function to the original input signal, the window function being of length 2N-1. The window function is non-rectangular, preferably triangular. Using a triangular window causes weighted window coefficients to be applied to the sample values, the ratio of the weighted window coefficients effecting the averaging process.

Figure 6B shows a series of 2N-1 weighted window coefficients w[n] for the example where N=4, a respective weighted window coefficient being provided for each input sample value x[n]. The value of each windowing coefficient may be determined based upon N, for example as shown in Figure 6B, which also shows specific window coefficient values for the case where N=4. The relative values of the window coefficients are such that the coefficient for the centre sample has the highest value (typically 1) and the respective coefficients for the other sample values on each side of the centre sample are progressively smaller. Hence, applying the weighted window coefficients to the respective sample values applies a triangular window function to the sample values that is centred on the centre sample (i.e. the highest value, or “apex”, of the triangle is applied to the centre sample). It is preferred that the triangular window is symmetrical about its apex, i.e. the window coefficients values are the same at corresponding positions on either side of the centre coefficient.

The window coefficient values can be implemented in hardware or software and can pre-stored in a system with a fixed N value or generated when needed. The input signal is windowed by multiplying each of the data samples by the respective window coefficient. The resulting weighted data samples are then paired together in a specific order, and each sample in each pair added together, to create an ApFFT averaged time series data vector. The pairing preferably involves pairing the or each data sample before the centre sample with a respective corresponding data sample after the centre sample, most preferable in time order, i.e. pairing the first data sample in the vector with the first data sample after the centre data sample, and the second data sample in the vector with the second data sample after the centre data sample, and so on. The samples in the averaged data vector are also ordered so that the centre sample is first and the summed pairs of data values follow in time order. The windowing method allows the same averaged data vector of N sample values to be created as the method described with reference to Figures 5A to 5C.

In any event, the 2N-1 sample input signal is pre-processed to create a single data vector to which a single N point FFT is applied, where the first sample of the input to the N point FFT is the Nth sample of the original time series of input data samples, and the other data samples in the single data vector are indicative of the entire input signal of 2N-1 samples.

Preferably, said identifying comprises selecting as said identified frequency transform value a peak frequency transform value of said frequency transform values corresponding to a frequency between 0 Hz and half of the sampling frequency.

In preferred embodiments, performing said frequency transform involves performing a Fast Fourier Transform (FFT) on at least some of said sampled output values, or on a plurality of values derived from said sampled output values. It is particularly preferred that performing said frequency transform involves performing an All Phase Fast Fourier Transform on a plurality of values derived from said sampled output values.

The preferred method further includes pre-processing said time sampled output values to produce a plurality of processed time values; and performing an FFT on said processed time values to produce said frequency transform values. Said pre-processing typically involves selecting one of said time sampled output values as a centre sample value, and arranging said time sampled output values, or a plurality of data values derived therefrom, into at least one series of data values in which said centre sample value is the first in the, or each, series. Said sampling typically involves producing 2N-1 sampled output values in a time series and said centre sample value is the Nth sample value in said time series.

It is preferred that said at least one series of data values in which said centre sample value is the first in the, or each, series has N data values including said centre sample value.

Preferably said arranging involves providing a single series of data values in which said centre sample value is the first in the series. Preferably said single series of data values comprises said centre sample value and at least one other data value comprising a weighted average of at least two other of said sampled output values. Preferably said single series of data values comprises said centre sample value and a plurality of other data values, each other data value comprising a weighted average of a respective two other of said sampled output values.

In preferred embodiments said pre-processing involves applying a triangular window function to said time sampled output values centred on said centre sample value to produce windowed sample values, and providing said single series of data values with said centre value as the first in the series, and at least one other data value comprising a summation of at least two of said windowed sample values. Preferably said single series of data values comprises said centre sample value and a plurality of other data values, each other data value comprising a summation of a respective two said windowed sample values.

Typically said FFT is applied to said single series of data values, and said FFT is usually an N-point FFT. detecting the lowest positive frequency at which frequency domain data, i.e. the magnitude of the FFT value, crosses the threshold, where the threshold is set to exclude FFT values that correspond to unwanted signal components such as noise. The detected FFT value, which is stored as a complex number, may be deemed to correspond to the fundamental frequency of the shock sensor output and, together with the frequency to which it corresponds, contains sufficient information to enable the phase of the shock sensor output to be determined. Other information such as a time corresponding to the occurrence of the detected peak may be obtained. This information may be used to make auto-location decisions, e.g. using the WAL or PAL technique, without the need for noise filtering.

In the example of Figure 8A, it can be seen that the relevant peak in the FFT magnitude occurs at 1Hz. The phase of the fundamental component of the input signal at the centre sample of the original input time series of data samples is then determined by examining the phase at the peak frequency, which in this case can be seen to be +70°.

The output of the ApFFT is stored as a complex number. The phase of the centre sample corresponds to the phase, or argument, of the complex ApFFT output value, and can be determined through trigonometry based on the relationship between the real and imaginary parts of the output values as shown in Figure 7. The phase, or argument, a is the inverse tan of the imaginary value divided by the real value.

Using the ApFFT as described above provides an accurate measure of centre sample phase making it ideal for use in WAL auto-location, or similar auto-location techniques, where the phase difference between two shock sensor signals is used to determine clockwise or anticlockwise wheel rotation. The foregoing processing can be performed on each shock sensor output signal and the results, or resulting determination on rotation direction, may be transmitted to the ECU for use in auto-location.

However, the frequency measurement step size is limited to the sampling frequency divided by the FFT size, also known as the frequency bin size. In the illustrated example, the bin size is 1Hz, resulting in a measured frequency of 1Hz, even though the input signal was actually 1.1 Hz. For this reason, this technique is not well suited for PAL auto-location, or similar auto-location techniques, where an accurate measure of both signal phase and frequency are required.

The signal frequency (i.e. the frequency of the fundamental frequency component of the shock sensor output in this case), as well as its phase at any sample in the data sequence, can be determined by using a technique called the All-Phase Difference method (DApFFT). Here, the ApFFT is used, in the manner described above, to measure the input time series phase at two points in time, spaced N samples apart. The difference in phase can then be used to determine the exact signal frequency. If the phase is known at two points in time, and the time between the points is known and the frequency is known to the nearest bin it is possible to determine the exact frequency of the signal.

This is illustrated in Figure 6A where N = 4 and there are 7 original time samples x[0] to x[6] spaced apart in time. The time spacing is illustrated by a respective delay element z'^ between the samples, which may be representative of an inter-sample delay corresponding to the sampling frequency, or may be implemented as a circuit delay element in a pipelined processing implementation. The window coefficients represented as elements w[0] to w[6] are generated, preferably as indicated in Figure 6B, and multiplied with the respective time samples x[0] to x[6]. The centre sample is x[3] and corresponding pairs of samples on either side of x[3] are added together (e.g. x[4] with x[0], x[5] with x[1] and x[6] with x[2]). Figure 6C shows the resulting time sample series, or data vector, comprising data samples Xap[0] to Xap[3], where Xap[0] is the windowed centre sample value and Xap[1] to Xap[3] are the paired and summed other windowed sample values. It can be seen that the data vector shown in Figure 6C is the same as the vector shown in Figure 5E, each having the centre sample value as the first value in the series followed by values that are weighted averages of respective pairs of the other original output values.

An N-point FFT is applied to the resulting N-point time series of data values (data vector) to produce an N-point FFT output. In Figure 6A this produces a 4 point FFT output comprising Xap[0] to Xap[3], Application of the N-point FFT to the N-point time series results in the phase of the original time series at the Nth sample, i.e. sample N-1, i.e. at the centre sample. In the example of Figure 6A, the 4-point FFT calculates the phase of the original 7-point time series at the 4^^ sample. It will be seen that the ApFFT generates accurate phase information but the frequency is accurate only to the closest bin.

By way of example, consider the following discrete-time sampled (with sampling frequency Fs = 6Hz) sinusoid with frequency (fo) 1.1 Hz and initial phase (po) of+100°:

The phase of the centre sample is

Forming, by way of example, a 6-point all-phase time series in the manner described above and then performing a 6-point FFT on the 6-point all-phase time series results in the frequency domain output illustrated in Figures 8A and 8B. Figure 8A shows the magnitude of the ApFFT output values and Figure 8B shows the corresponding phase for each value.

The phase of the fundamental frequency component is found by first identifying the fundamental frequency component between OHz and Fs/2 Hz. Identifying the fundamental frequency component may be achieved by identifying a peak value in the FFT values that may be deemed to correspond to the fundamental frequency of the shock sensor output. This may involve setting a threshold, and

For an N-point DApFFT, a time series comprising 3N-1 time samples (i.e. of the sampled sensor output signal) are taken and spit into two overlapping segments, each segment comprising a time series with 2N-1 time samples of the original 3N-1 samples. The two segments are overlapping in that they are spaced apart by N samples, and so N-1 of the time samples appear in both segments. The N-point ApFFT is applied to each segment in the manner described above, and the results are used to identify the fundamental frequency to the nearest bin and to determine the phase at the centre sample of each segment in the manner described above. The difference between the phase values for each segment can be used to correct the frequency of the fundamental frequency component as determined using the ApFFT. The corrected frequency can then be used to accurately calculate the phase of the sampled shock sensor output signal at any sample in the original time sample series.

For phase difference (p<jiff) in the range 4;^: radians, the ratio of phase difference to 2π is the same as the ratio of frequency difference (fdiff) to the bin size (Fs/N):

The ratio of phase difference to 2® is called the leakage error coefficient and is given the symbol, d:

So, the frequency difference as a fraction of the frequency bin size is given by:

The corrected frequency estimate, may be found by combining the ApFFT frequency estimate (fApFFi) with the frequency difference:

Finally, the phase per sample, is equal to:

Knowing both the signal phase at sample N-1 and the phase per sample allows the estimation of input signal phase at any sample in the input time series, using linear interpolation.

Once the shock sensor phase at the required sample has been determined, this information can be correlated to the ABS tooth count data to achieve auto-location, according to existing PAL algorithms or similar algorithms.

Once the relevant information is available (phase and frequency for PAL, clockv\/ise or anticlockwise rotation for WAL or any other information for any other strategy), the relevant information is transmitted to the ECU 300.

The invention is not limited to the embodiment(s) described herein but can be amended or modified without departing from the scope of the present invention.

Claims

Claims:

1. A method of processing an output of a wheel phase angle sensor of a wheel monitoring unit mounted on a wheel to determine a phase value for said wheel phase angle sensor, the method comprising: sampling, when said wheel is rolling, said output to produce time sampled output values; storing a plurality of said time sampled output values; performing a frequency transform on said stored output values, or a plurality of values derived therefrom, to produce a plurality of frequency transform values corresponding to respective frequency components of said wheel phase angle sensor output; identifying one of said frequency transform values as corresponding to a fundamental frequency component of said wheel phase angle sensor output; and determining said phase value from the identified frequency transform value.

2. The method of claim 1, wherein said frequency transform values are represented as a respective complex number, said determining said phase value comprising calculating the phase of the complex number corresponding to said identified frequency transform value.

3. The method of claim 1 or 2, wherein said identifying comprises selecting as said identified frequency transform value a peak frequency transform value of said frequency transform values corresponding to a frequency between 0 Hz and half of the sampling frequency.

4. The method of any preceding claim, wherein performing said frequency transform involves performing a Fast Fourier Transform (FFT) on at least some of said sampled output values, or on a plurality of values derived from said sampled output values.

5. The method of claim 4, wherein performing said frequency transform involves performing an All Phase Fast Fourier Transform on a plurality of values derived from said sampled output values.

6. The method of claim 4 or 5, further including pre-processing said time sampled output values to produce a plurality of processed time values; and performing an FFT on said processed time values to produce said frequency transform values.

7. The method of claim 6, wherein said pre-processing involves selecting one of said time sampled output values as a centre sample value, and arranging said time sampled output values, or a plurality of data values derived therefrom, into at least one series of data values in which said centre sample value is the first in the, or each, series.

8. The method of claim 7, wherein said sampling involves producing 2N-1 sampled output values in a time series and said centre sample value is the Nth sample value in said time series.

9. The method of claim 8, wherein said at least one series of data values in which said centre sample value is the first in the, or each, series has N data values including said centre sample value.

10. The method of any one of claims 7 to 9, wherein said arranging involves providing a single series of data values in which said centre sample value is the first in the series.

11. The method of claim 10, wherein said single series of data values comprises said centre sample value and at least one other data value comprising a weighted average of at least two other of said sampled output values.

12. The method of claim 10 or 11, wherein said single series of data values comprises said centre sample value and a plurality of other data values, each other data value comprising a weighted average of a respective two other of said sampled output values.

13. The method of any one of claims 10 to 12, wherein said pre-processing involves applying a triangular window function to said time sampled output values centred on said centre sample value to produce windowed sample values, and providing said single series of data values with said centre value as the first in the series, and at least one other data value comprising a summation of at least two of said windowed sample values.

14. The method of claim 13, wherein said single series of data values comprises said centre sample value and a plurality of other data values, each other data value comprising a summation of a respective two said windowed sample values.

15. The method of any one of claims 10 to 14, wherein said FFT is applied to said single series of data values.

16. The method of claim 15 when dependent on claim 8 wherein said FFT is an N-point FFT.

17. The method of any one of claims 7 to 9, wherein said arranging involves providing a plurality of series of data values in which said centre sample value is the first in each series, said FFT being applied to each series to produce a respective set of frequency coefficient values, said frequency coefficient values being averaged to produce said frequency transform values.

18. The method of claim 17 when dependent on claim 8, wherein N series of data values are provided, each comprising N data values including the centre sample value, and said FFT is an N-point FFT.

19. The method of claim 17 or 18, wherein each series of data values contains a respective set of said time sampled output values, the sets being time shifted with respect to each other and overlapping such that each set includes the centre sample value, and each other time sampled output value is included in at least one set.

20. The method of any preceding claim, further including determining a value for the fundamental frequency of said wheel phase angle sensor output by: determining an estimate of said fundamental frequency value from said identified frequency transform value; determining the respective phase value for first and second sets of time sampled output values, the sets being spaced apart in time by at least one sample period; calculating a frequency correction value from the difference between the respective phase values; and adjusting said estimate of said fundamental frequency value using said frequency correction value to calculate said fundamental frequency value.

21. The method of claim 20, wherein said first and second sets of time sampled output values are spaced apart by N samples, and wherein calculating said frequency correction value involves multiplying the phase difference by Fs/(2πN), where Fs is the sampling frequency.

22. The method of claim 21, further including calculating a phase per sample value by multiplying the calculated fundamental frequency value by π/Fs, and determining a phase value for any sample value of the first or second time sampled output values using said phase per sample value and linear extrapolation.

23. A method of performing wheel auto-location, the method including the method as claimed in any preceding claim.

24. The method of claim 23, including processing a respective output of first and second wheel phase angle sensors of a wheel monitoring unit using the method as claimed in any one of claims 1 to 22.

25. The method of claim 23, including processing an output of a single wheel phase angle sensor of a wheel monitoring unit to calculate the phase value and fundamental frequency value using the method of any one of claims 20 to 22.

26. Awheel monitoring unit comprising: at least one wheel phase angle sensor, the method comprising: means for sampling, when said wheel is rolling, said output to produce time sampled output values; means for storing a plurality of said time sampled output values; means for performing a frequency transform on said stored output values, or an a plurality of values derived therefrom, to produce a plurality of frequency transform values corresponding to respective frequency components of said wheel phase angle sensor output; means for identifying one of said frequency transform values as corresponding to a fundamental frequency component of said wheel phase angle sensor output; and means for determining said phase value from the identified frequency transform value.

27. The wheel monitoring unit as claimed in claim 26, comprising means for performing the steps of any one of the methods of claims 1 to 25.

28. A tyre pressure monitoring system comprising at least one wheel monitoring unit as claimed in claim 26 or 27, preferably a respective wheel monitoring unit for each wheel of a vehicle.