JP2006224757A - Wheel abnormality detecting device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a wheel abnormality detecting device capable of detecting abnormality about all wheels even while a vehicle travels at high speed. <P>SOLUTION: The wheel abnormality detecting device 10 comprises a vibration measuring means 1 continuously measuring the vibration strength of each wheel, a wheel speed measuring means 2 continuously measuring the rotation speed of each wheel, an extracting means 3 extracting a waveform synchronizing with rotations of the wheel from data measured by the vibration measuring means and the wheel speed measuring means, and the warning means 4 judging a state of the wheel from the waveform extracted by the extracting means 3 and warning abnormality when the judged state is out of a predetermined normal range, provided in the plurality of wheels equipped to a vehicle during traveling. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an apparatus for detecting an abnormality of a wheel mounted on a traveling vehicle, and more particularly to an apparatus for improving a processing speed required for detection.

  As a device for detecting an abnormality of a wheel, for example, a tire constituting the wheel, an internal pressure abnormality alarm device is known that issues an alarm and warns a driver when the tire internal pressure becomes a predetermined value or less. In this apparatus, an internal pressure is measured by an internal pressure sensor attached to a wheel of a tire, and a warning is issued when the pressure falls below a certain internal pressure.

  However, this internal pressure abnormality alarm device cannot detect a failure caused by a cause other than a decrease in internal pressure. For example, as an example of such a failure, separation between the tread and the belt, between the cords constituting the belt, between the side rubber and the carcass ply, breakage of the ply cord or the belt cord, and chunk-out of the tread rubber ( For example, there is a state in which the block land portion provided in the tread is torn off. If the vehicle continues to run in a state where these tire failures occur, a sudden tire burst may occur and the vehicle may become unable to run, and in addition, a serious accident may occur.

As a system for detecting failures other than such internal pressure, the system for detecting abnormalities in the rotating body such as tire abnormalities is measured by measuring tire vibration and sound data and comparing them with normal data obtained in advance. Although known (see, for example, Patent Document 1), even with this abnormality detection system, it is possible to detect a tire abnormality. However, in recent years, there is a demand for a vehicle abnormality detection system with a simpler configuration and higher performance. It was.
JP 2003-80912 A

  Under such a background, the present applicant, for example, continuously measures the vibration of the wheel, and uses an adaptive digital filter to detect only the waveform synchronized with the rotation of the wheel when detecting an abnormal state from the vibration data. We have proposed a device that detects wheel abnormalities by taking out the motor and comparing the waveforms with normal ones.

  However, although this proposal can detect the abnormality of the wheels with high accuracy in principle, the detection of the abnormality of a plurality of wheels attached to the vehicle in a short time in a vehicle traveling at a high speed can be detected. In order to do it frequently enough, further contrivances were required.

  The present invention has been made in view of such problems, and in a vehicle in which a plurality of wheels are mounted, even if the vehicle is traveling at a high speed, the wheel is frequently Another object of the present invention is to provide a wheel abnormality detection device capable of checking the abnormality and detecting the abnormality.

<1> is the vibration measuring means for continuously measuring the vibration intensity of each wheel with respect to a plurality of wheels mounted on the running vehicle, and the rotational speed of each wheel directly or A wheel speed measuring means for continuously measuring the vehicle characteristics correlated with the speed, and a waveform synchronized with the rotation of the wheel from the data measured by the vibration measuring means and the wheel speed measuring means. In the wheel abnormality detection device comprising the extracting means for extracting and the abnormality warning means for determining the state of the wheel from the waveform extracted by the extracting means and warning the abnormality when this state is out of the predetermined normal range,
The extraction means includes
Fixed to generate a “measurement waveform” consisting of a data sequence arranged in time order by sampling vibration intensity data for each wheel continuously measured by the vibration measuring means at a predetermined fixed time interval δ _f A sampling processing unit;
For each wheel, only the component synchronized with the rotation of the wheel is extracted from the measurement waveform as a data sequence arranged in time order, and this is output as the “synchronous component waveform” to the abnormality warning means. A filter section;
It consists of a model waveform generation unit that generates a “model waveform” as a model of the extraction of each wheel as a data sequence arranged in time order,
The adaptive digital filter unit receives the model waveform, generates the synchronization component waveform based on both the difference between the measurement waveform and the synchronization component waveform, and the input model waveform, and the synchronization component waveform Is configured to adaptively control the filtering parameter in generating the difference using the difference,
When the frequency obtained by multiplying the fundamental frequency corresponding to the period of one rotation of each wheel by a natural number n is the `` nth synchronization frequency '' for that wheel,
The model waveform for each wheel, frequency domain waveform of time domain waveform obtained by arranging the data in the time interval of the [delta] _f can be the Fourier transform of this model waveform, the wheel, the next number of the synchronization frequency according to the extracted It becomes a mountain at the position of, and forms a valley between adjacent synchronization frequencies of the order, the top of the portion that becomes the mountain extends to a frequency range of infinitesimal or finite width as a point or flat surface, The height of the top is configured to have the same value for all orders of synchronization frequency for the extraction,
The model waveform generation unit sets a basic frequency of each wheel or a range including the basic frequency based on data from the wheel speed measuring unit, and sets the frequency domain waveform characteristics from the set basic frequency or the range. Is a wheel abnormality detection device configured to generate a model waveform that satisfies

  Here, “wheel speed measuring means for continuously and continuously measuring the rotational speed of each wheel by measuring the vehicle characteristics correlated with the wheel speed” means, for example, the vehicle characteristics correlated with the wheel speed. The vehicle speed sensor in the case of calculating the range of the wheel speed of each wheel by measuring the vehicle speed.

<2> in <1>, wherein the adaptive digital filter unit is configured to generate respective synchronous component waveforms for a plurality of wheels based on a single model waveform common to these,
In the frequency domain waveform corresponding to the common model waveform, the frequency range of the peak corresponding to the n-th synchronization frequency is the n-th order for all wheels related to the extraction of the synchronization component waveform based on the common model waveform. It is a wheel abnormality detection apparatus characterized by including these synchronous frequencies.

  <3> is the extraction of the synchronization component waveform based on the common model waveform, with the frequency range at the peak corresponding to the n-th synchronization frequency being defined as a predetermined percentage in <2>. Among the plurality of wheels, the wheel abnormality detection device is set as 0 to + A% of the n-th order synchronization frequency of the wheel having the lowest fundamental frequency.

  <4> is the fundamental frequency among the plurality of wheels according to <2>, in which the frequency range of the peak corresponding to the n-th synchronization frequency is extracted from the synchronous component waveform based on the common model waveform Is a wheel abnormality detection device that is set on the basis of the n-th synchronization frequency for both wheels, the wheel having the smallest and the wheel having the largest fundamental frequency.

  <5> is any one of <2> to <4>, wherein the adaptive digital filter unit is based on the common model waveform every time one piece of data constituting the common model waveform is input. The wheel abnormality detection device is configured to perform a process of outputting only one piece of data of the synchronous component waveform for each of all the wheels related to the extraction of the synchronous component waveform.

<6> is the vibration measuring means for continuously measuring the vibration intensity of each wheel with respect to a plurality of wheels mounted on the running vehicle, and the rotational speed of each wheel directly or the wheel. A wheel speed measuring means for continuously measuring the vehicle characteristics correlated with the speed, and a waveform synchronized with the rotation of the wheel from the data measured by the vibration measuring means and the wheel speed measuring means. In the wheel abnormality detection device comprising the extracting means for extracting and the abnormality warning means for determining the state of the wheel from the waveform extracted by the extracting means and warning the abnormality when this state is out of the predetermined normal range,
The extraction means includes
For each wheel, the vibration intensity data continuously measured along the time axis by the vibration measuring means is sampled at a time interval δ _v that changes according to the data from the wheel speed measuring means, and the data is arranged in time order. A variable sampling processing unit that generates a "measurement waveform" consisting of columns;
For each wheel, only the component synchronized with the rotation of the wheel is extracted from the measurement waveform as a data sequence arranged in time order, and this is output as the “synchronous component waveform” to the abnormality warning means. A filter section;
It consists of a model waveform generation unit that generates a “model waveform” as a model of the extraction of each wheel as a data sequence arranged in time order,
The time interval δ _v used to generate the measurement waveform for each wheel in the variable sampling processing unit corresponds to a cycle of one rotation of the wheel calculated based on data from the wheel speed measuring means. Or a fundamental frequency f _a corresponding to a cycle of one rotation of the virtual wheel calculated by a predetermined procedure based on data from the wheel speed measuring means and one rotation of these wheels. It is obtained as the reciprocal of the product with a predetermined value k which is the number of divisions when equally dividing,
The adaptive digital filter unit receives the model waveform, generates the synchronization component waveform based on both the difference between the measurement waveform and the synchronization component waveform, and the input model waveform, and the synchronization component waveform The filtering parameter when generating is adaptively controlled using the difference, and each synchronous component waveform for a plurality of wheels is configured to be generated based on one model waveform common to these wheels,
A predetermined fixed frequency is set as a fixed basic frequency f ₀ , and a frequency obtained by multiplying the fixed basic frequency by a natural number n is referred to as an “n-th fixed synchronization frequency”, and the fixed basic frequency f ₀ and the division number k are When the reciprocal of the product of the time interval δ ₀ ,
The common model waveform is a frequency domain waveform obtained by Fourier transforming a time domain waveform in which each data of the model waveform is arranged at every time interval δ ₀ , and is a peak at the position of the fixed synchronization frequency of the order related to the extraction. It forms a valley between the fixed synchronization frequencies of adjacent orders, and the top of the peak portion extends as a point or flat surface in a frequency range of infinitesimal or finite width, and the height of the top The wheel abnormality detection device is configured to have the same value with respect to the fixed synchronization frequencies of all orders related to the extraction.

  Here, “wheel speed measuring means for continuously and continuously measuring the rotational speed of each wheel by measuring the vehicle characteristics correlated with the wheel speed” means, for example, the vehicle characteristics correlated with the wheel speed. The vehicle speed sensor in the case of calculating the range of the wheel speed of each wheel by measuring the vehicle speed.

  In addition, “virtual wheel” in “basic frequency corresponding to a cycle of one rotation of a virtual wheel calculated by a predetermined procedure based on data from the wheel speed measuring means” is, for example, a measurement waveform When a wheel that is different from the wheel to be generated is a “virtual wheel”, or a wheel with a predetermined diameter that travels at the same speed as the vehicle speed while rolling on the road surface with a slip rate of zero is a “virtual wheel” The “virtual wheel” in this case.

<7> in <6>, the time interval δ _v used for the sampling process corresponding to each wheel is the same for all of the plurality of wheels related to the extraction of the synchronous component waveform, Of the fundamental frequencies of these wheels, the minimum frequency is f _min , the maximum frequency is f _max , and the frequency used for calculating the time interval δ _v is f _a, and the frequency domain waveform corresponding to the model waveform , The peak frequency range corresponding to the n-th fixed synchronization frequency is at least n times f ₀ · (f _min / f _a ) and n times f ₀ · (f _max / f _a ) The wheel abnormality detection device is configured to include the following ranges.

<8>, in <7>, of the basic frequency for all of the wheels according to the extraction of the synchronizing component waveform based on said common model waveform, the smallest ones, the, used to calculate the common time interval [delta] _v In addition, the wheel abnormality is defined by setting the frequency range of the peak corresponding to the n-th fixed synchronization frequency as a range of 0 to + A% of the n-th fixed synchronization frequency, where A is a predetermined percentage. It is a detection device.

  <9> is the peak frequency range of the plurality of wheels related to extraction of the synchronous component waveform based on the common model waveform in the frequency range of the peak corresponding to the n-th fixed synchronous frequency in <7>. This is a wheel abnormality detection device that is set based on both n-th order synchronization frequencies of the smallest wheel and the wheel having the maximum fundamental frequency.

  <10> is any one of <7> to <9>, wherein the adaptive digital filter unit synchronizes based on the common model waveform every time one piece of data constituting the common model waveform is input. This is a wheel abnormality detection device configured to perform processing for outputting only one piece of synchronous component waveform data for each of all the wheels related to component waveform extraction.

According to <1>, vibration intensity data is generated by sampling at predetermined time intervals, and a “measurement waveform” composed of a data sequence arranged in time order and a “model waveform” composed of a data sequence arranged in time order, Since the synchronous component waveform that is synchronized with the rotation of the wheel is extracted, it is possible to process the samples sampled at the predetermined time interval δ _f as they are, without performing time-consuming processing such as Fourier transformation. Thus, one extraction process can be shortened, and even if the vehicle is in a high-speed running state, abnormality check can be performed with sufficient frequency for all of the wheels.

  According to <2>, the adaptive digital filter unit is configured so that the respective synchronous component waveforms for the plurality of wheels are generated based on a single model waveform that is common to the plurality of wheels. Time can be reduced and the synchronous component waveform extraction process can be further shortened.

  According to <3>, the frequency range of the peak corresponding to the n-th synchronization frequency in the frequency domain waveform corresponding to the model waveform is set based on only the fundamental frequency of the fastest wheel. As will be described later, the apparatus can be made cheaper and simpler.

  According to <4>, in the frequency domain waveform corresponding to the model waveform, the frequency range of the top of the mountain corresponding to the n-th synchronization frequency is defined as the basic frequency of the fastest wheel and the basic frequency of the slowest wheel. Therefore, the frequency components synchronized with the wheels can be reliably extracted as will be described in detail later.

  According to <5>, the adaptive digital filter unit inputs each piece of data constituting the common model waveform for each of the wheels related to extraction of the synchronous component waveform based on the common model waveform. Since it is configured to perform processing to output only one piece of synchronous component waveform data, in order to generate the i-th synchronous component waveform data, a call from the model waveform memory or buffer is made for each wheel. This can be done only once for all wheels, and the synchronization component waveform extraction process can be further shortened.

  According to <6>, since a “measurement waveform” composed of a data sequence arranged in time order and a “model waveform” composed of a data sequence arranged in time order, a synchronous component waveform synchronized with the rotation of the wheel is extracted. Similar to <1>, it is possible to frequently check for abnormalities even if the vehicle is in a high-speed traveling state by shortening one extraction step.

  According to <7>, since the sampling in the variable sampling processing unit is configured to be performed based on one common time interval, the time taken to calculate the sampling time interval is reduced, and the synchronous component waveform The extraction process can be further shortened.

  According to <8>, since the frequency range of the peak corresponding to the n-th synchronization frequency in the frequency domain waveform corresponding to the model waveform is set based only on the fundamental frequency of the fastest wheel, <3 As with>, the device can be made cheaper and simpler.

  <9> According to the frequency domain waveform corresponding to the model waveform, the frequency range of the top of the mountain corresponding to the n-th synchronization frequency is expressed by the basic frequency of the fastest wheel and the basic frequency of the slowest wheel. Therefore, it is possible to reliably extract the frequency component synchronized with the wheel as in <4>.

  According to <10>, the adaptive digital filter unit inputs each piece of data constituting the common model waveform for each of the wheels related to extraction of the synchronous component waveform based on the common model waveform. Since it is configured to output only one piece of synchronous component waveform data, as in <5>, the model waveform memory or buffer is not called for each wheel. However, the process of extracting the synchronous component waveform can be further shortened.

As embodiments of the present invention, first to fourth embodiments will be described below with reference to the drawings. The meanings of the main symbols used in these explanations are as follows.
<Explanation of symbols>
r: Wheel number (r = 1 to R, R is the number of wheels)
i: Data order in the data sequence along the time axis (i = 1 to M, M is the number of samplings)
p: Embodiment number (p = 1 to 4)
a _r : R wheel vibration intensity data
v _r : Wheel speed strength data of the r-th wheel
f _r ⁽ⁿ⁾ : n-th synchronization frequency of the r-th wheel (f _r ⁽¹⁾ is the fundamental frequency)
X _pr , X _pr (i): measurement waveform of the r-th wheel and its i-th data in the p-th embodiment
Y _pr , Y _pr (i): the synchronous component waveform of the r-th wheel in the p-th embodiment and its i-th data
R _pr , R _pr (i): model waveform of the r-th wheel in the p-th embodiment and its i-th data
R _p , R _p (i): model waveform common to each wheel in the p-th embodiment and its i-th data
E _pr , E _pr (i): error of the r-th wheel in the p-th embodiment and its i-th data δ _f : time interval in fixed sampling δ _vr : in variable sampling based on wheel speed data of the r-th wheel Time interval

  Fig.1 (a) is a plane schematic diagram which shows arrangement | positioning of the wheel abnormality detection apparatus in a vehicle, FIG.1 (b) is a side surface schematic diagram of the wheel which shows the abnormal part which arose in a part of wheel, FIG. 2 is a block diagram showing the configuration of the wheel abnormality detection device.

The wheel abnormality detection apparatus 10 includes a vibration measuring unit 1 that continuously measures the vibration intensity of each wheel with respect to the plurality of wheels 90A, 90B, 90C, and 90D mounted on the traveling vehicle, and the rotation of each wheel. A wheel speed measuring means 2 for continuously measuring the speed; an extracting means 3 for extracting a waveform synchronized with the rotation of the wheel from the data measured by the vibration measuring means 1 and the wheel speed measuring means 2; An abnormality warning means 4 for determining the state of the wheel from the extracted waveform and warning the abnormality when the state is out of a predetermined normal range is provided. When the vehicle is, for example, a four-wheeled vehicle, vibration measurement is performed. Means 1 is composed of first to fourth vibration sensors 1a to 4d attached to the knuckles that support the wheels 90A to 90D, for example, and these vibration sensors 1a to 4a are That , Continuously outputs the vibration intensity data a ₁ ~a ₄ to the extraction unit 3.

When the vehicle is a four-wheeled vehicle, the wheel speed measuring means 2 includes wheel speed sensors 2a to 2d for the first to fourth wheels that detect the rotational speed of each wheel, and these wheel speed sensors 2a to 2d. Respectively output the wheel speed data v _{1 to} v ₄ to the extraction means 3 continuously. Examples of the vibration sensors 1a to 4d include acceleration sensors that detect the acceleration in the longitudinal direction of the vehicle. Examples of the wheel speed sensors 2a to 2d are attached to the respective axles to control the ABS. The thing using the used gear pulsar can be mentioned.

Extracting means 3, from each of the waveform consisting of vibration intensity data a ₁ ~a _4, and extracts only a waveform b ₁ ~b ₄ synchronized with the rotation of the wheels, and outputs them to the abnormality warning means 4. The extraction unit 3 and the abnormality warning unit 4 are provided on the vehicle body side, and the abnormality warning unit 4 includes an abnormality control unit 4a that performs abnormality determination and an abnormality display unit 4b that warns the driver in the event of an abnormality. The extraction means 3 and the abnormality control unit 4a are arranged in the central control device 5.

  The hardware configuration of the central control device 5 can be configured by what is used for a general control device, and includes, for example, a CPU, a memory, an input / output interface including AD conversion, and the like.

  As shown in FIG. 1B, in a running vehicle, if any abnormality occurs in any of the wheels 90A, for example, delamination 91 occurs between the tire tread and the belt, a vibration waveform synchronized with the rotation of the wheels is generated. appear.

The wheel abnormality detection device 10 according to the present invention extracts only the vibration waveform synchronized with the rotation of the wheel from the vibration intensity data a ₁ in which vibrations from the road surface and the like are mixed, and the waveform exceeds a predetermined range. The vehicle is configured to have a function of warning the driver of any abnormality that has occurred in the vehicle.

First, a first embodiment of the vehicle abnormality detection device 10 of the present invention will be described. FIG. 3 is a block diagram showing, as an example, the extraction means 3A that constitutes a part of the vehicle abnormality detection device of this embodiment, which is provided in a four-wheeled vehicle. The vibration intensity data a _{1 to} a ₄ for each wheel continuously measured are sampled at a predetermined fixed time interval δ _f and measured waveforms X _{11 to} X ₁₄ consisting of a data sequence arranged in time order. a fixed sampling processing unit 11 for generating, for each wheel, from among the measured waveform X ₁₁ to X _14, only components synchronized with the rotation of the wheels, and extracted as a data string arranged in time order, this synchronization The adaptive digital filter unit 12 that outputs the component waveforms b _{1 to} b ₄ (or Y _{11 to} Y ₁₄ ) to the abnormality warning means 4 and the model waveforms R _{11 to} R ₁₄ that serve as the models for the extraction of each wheel, Generated as a sequence of data The more the model waveform generating unit 13 that.

The fixed sampling processing unit 11 includes processing units (1) to (4) corresponding to the wheels 90A to 90D. For example, in the processing unit (1), the vibration intensity illustrated by taking time on the horizontal axis in FIG. The data waveform a ₁ is sampled at a predetermined time interval δ _f that is unrelated to the wheel speed v _1, and is sampled as shown in the graph of FIG. 5 with the horizontal axis representing the data order along the time axis. the data and outputs measurement waveform X ₁₁ which time-sequenced. In FIG. 5, X ₁₁ (i-1), X ₁₁ (i), and X ₁₁ (i + 1) are respectively arranged in a time sequence continuously to form a data string, i-1, i, i + The first data is shown.

The sampling number M for one extraction step uses a predetermined value, but includes 2 to 5 rotations of the wheel in order to make the extraction process using the adaptive digital filter have sufficient accuracy. At the same time, the division number k representing the sampling number for one round of the tire is preferably 360 to 3600, and therefore the sampling number M is preferably 720 to 18000. At this time, the fixed sampling time interval δ _f is preferably set to 20 to 200 μs with a design speed of 100 km / h and a wheel circumference of about 2 m.

The adaptive digital filter unit 12 includes an adaptive digital filter, ADF (1) to (4), provided for each wheel. Figure 6 shows, as an example, is a graph showing a portion of a synchronous component waveform Y ₁₁ output from ADF (1), taking data order along the horizontal axis a time axis, in FIG. 6, Y ₁₁ (i- 1), Y ₁₁ (i), and Y ₁₁ (i + 1) respectively indicate i−1, i, and i + 1th data arranged sequentially in time order in the data string.

Each ADF, for example, the first wheel of the ADF (1) receives the corresponding model waveform R _11, a difference between the measured waveform X ₁₁ synchronous component waveform Y ₁₁ as E _11, model waveform type differential E ₁₁ to generate a synchronizing component waveform Y ₁₁ on the basis of the R _11, the filtering parameters in generating the synchronizing component waveform Y _11, configured to adaptively controlled using the difference E _11.

Here, as a method of optimizing by feeding back the difference E _1r to the filtering parameter, adaptation such as the LMS (Least Mean Square) method, Newton method, or the steepest bottom method, which is conventionally known as a filtering parameter update algorithm, is used. Algorithms that can be used, and other adaptive algorithms that can be suitably used include Complex LMS Algorithm (Complex Least Mean Square Algorithm), Normalized LMS Algorithm (Normalized Least Mean Square Algorithm), Projection Algorithm, SHARF Algorithm ( Simple Hyperstable Adaptive Recursive Filter Algorithm), RLS algorithm (Recursive Least Square Algorithm), FLMS algorithm (Fast Least Mean Square Algorithm), adaptive filter using DCT (Adaptive Filter using Discrete Cosine Transform), SAN filter (Single F requency Adaptive Notch Filter), neural network, and genetic algorithm can be used.

Further, the model waveform R ₁₁ to R _14, from the measured waveform X ₁₁ to X _14, when extracting only a component synchronized with the rotation of the wheel, functions as a cutting die used to punch the portion to be extracted The details of this waveform will be described below. In the following description, the frequency corresponding to the period of one wheel rotation is called a fundamental frequency, and the fundamental frequency is multiplied by a natural number n including 1. Is referred to as the nth order synchronization frequency for that wheel.

Each of the model waveforms R _{11 to} R ₁₄ is generated by the generation units (1) to (4) corresponding to the wheels of the model waveform generation unit 13. FIG. 7 is a graph showing, as an example, a part of the synchronous component waveform R ₁₁ generated by the generation unit (1) for the first wheel, with the horizontal axis representing the data order along the time axis. ₁₁ (i-1), R ₁₁ (i), and R ₁₁ (i + 1) respectively indicate i-1, i, and i + 1th data arranged sequentially in time order in the data string. .

The data R ₁₁ (i) to R ₁₄ (i) of the model waveforms R _{11 to} R ₁₄ for each wheel can be arranged at a fixed time interval δ _f used for sampling the vibration intensity data a _{1 to} a _4. Fourier transform the time domain waveform, the frequency domain waveform, Figure 8, those corresponding to the model waveform R _11, as illustrated frequency abscissa, the following number of synchronization frequency according to the extracted synchronization component waveform Y ₁₁ It becomes a mountain at the position of, and forms a valley between adjacent synchronization frequencies of the order, the top of the portion that becomes the mountain extends to a frequency range of infinitesimal or finite width as a point or flat surface, The height of the top has the same value with respect to the synchronization frequencies of all orders related to the extraction.

That is, in FIG. 8, the frequency waveform corresponding to the model waveform R ₁₁ of the wheel 90A is an integer multiple n (n = 1 ⁾ of the basic frequency f ₁ ⁽¹⁾ calculated based on the data from the wheel speed sensor of the wheel 90A. ~ 20) with a frequency f ₁ ⁽¹⁾ ~ f ₁ ⁽²⁰⁾ (where f ₁ ⁽ⁿ⁾ = n · f ₁ ⁽¹⁾ ) In the case of what is shown in FIG. 8, the top of the peak portion is arranged as an infinitely small point, and the height of the top is at all the positions of the frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ . The shape has the same value.

On the other hand, in order to generate the model waveform R ₁₁ that is a data sequence arranged in time order from the calculated fundamental frequency f ₁ ⁽¹⁾ , for example, the time domain waveform R ₁ (t) is calculated by Equation (1). Then, the calculated waveform may be sampled at a time interval δ _f to form a data string along the time axis. Here, N represents the maximum order related to the extraction, and A ₁ is a predetermined constant.

In order to confirm the success or failure of extraction of the component synchronized with the rotation of the wheel, the extraction means 3A configured as described above was sampled at a fixed time interval δ _f from the vibration intensity data a ₁ of the traveling wheel 90A. Each of the measurement waveform X ₁₁ and the synchronous component waveform Y ₁₁ output from the digital filter unit, and the data X ₁₁ (i) and Y ₁₁ (i) constituting these waveforms at the time interval δ _f A time domain waveform was generated side by side, and these were further Fourier transformed to obtain a frequency domain waveform. Figure 9 is a frequency domain waveform corresponding to the measured waveform X _11, FIG. 10 is a graph showing a frequency domain waveform corresponding to the synchronizing component waveform Y _11.

When FIG. 9 is compared with FIG. 10, FIG. 10 is composed of peaks having a height having the same value as the vibration intensity in FIG. 9 at the synchronization frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ . and forms a waveform in the form of troughs are formed in the waveform to synchronize the measurement waveform X ₁₁ to the rotation of the wheel, i.e., it indicates that it is possible to extract the waveform consisting synchronizing frequency of the wheel In the abnormality warning device 4, this height is determined in advance by performing processing for determining the height of the peaks at the synchronization frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ or processing equivalent thereto. When the limit value is exceeded, it can be configured to determine that some abnormality has occurred in the wheel.

The measurement waveform in FIG. 9 is based on a sampling process in which the time interval δ _f at the time of sampling is 140 μs and the sampling number M is 2048, and the fundamental frequency at the time of measurement is 13.5 Hz.

FIG. 11 is a flowchart showing the processing executed by the extraction means 3A of the first embodiment. The extraction means 3 first sets the wheel number r to 1 with the number of wheels set to 4 (step s101). For the wheels, wheel speed data v ₁ is input (step s102), then the basic frequency of the first wheel is calculated based on the wheel speed data v ₁ and the model waveform R ₁₁ (i) (i = 1 to M, However, M generates the sampling number used for one extraction) (step s103). For this generation, as in the procedure described above, the time domain waveform is obtained by using the equation (1), and this is discretized at the time interval δ _f to form a data string.

Then, calculate the i-th data Y ₁₁ synchronization component waveform Y ₁₁ from the i-th data X ₁₁ of the i-th data R ₁₁ (i) and the measured waveform X ₁₁ of model waveform R ₁₁ (i) (i) This follows the following procedure. That is, first, the synchronous component waveform is initialized (step s104), the data number is initialized to i = 1 (step s105), and then the model waveform data R ₁₁ ( 1) is input to the ADF (1) (step s106). On the other hand, the measurement waveform X is first sampled at the time interval δ _f while acquiring the vibration intensity data a ₁ (step s107). ₁₁ first data X ₁₁ (1) is generated (step s108), and difference E ₁₁ (1) is obtained from X ₁₁ (1) and initialized Y ₁₁ (0) (step s109). Input to ADF (1).

In ADF (1), together with processes optimized filtering parameters by the difference E ₁₁ (1), and R ₁₁ (1) entered previously, because the difference E ₁₁ (1), the first synchronization component waveform Y ₁₁ Data Y ₁₁ (1) is obtained (step s110), and is output for use in calculation of the next difference E ₁₁ (2) and is output to the abnormality warning device 4 (step s111).

With the above, one round of steps for obtaining Y ₁₁ (1) is completed, i = i + 1 is set (step s112), and Y ₁₁ (2) is entered, but before that, i is the sampling number M determines whether the reached (step s113), if they reached if completes the extraction of the sync component waveform Y ₁₁ for the first wheel, as r = r + 1 (step s114), the second wheel for, it was extracted synchronization component waveform Y ₁₁ in the same procedure, which to a final wheel (step s115) repeatedly Kaee to complete the single extraction step. And while driving | running | working, this extraction process is repeatedly performed with a predetermined period, The more real-time abnormality detection can be performed, so that this repetition is performed with a short period.

  In addition, although the processing for each wheel is assumed to be in series in time for convenience of explanation, it can be parallel processing as long as the processing time of the apparatus is allowed. In this case, one extraction step Can be shortened.

As described above, when calculating the synchronous component waveform Y _1r , the input vibration intensity data a _r is sampled at a predetermined time interval δ _f without performing time-consuming processing such as Fourier transform, processed input order to be able to generate a synchronizing component waveform Y _11, by this, without sacrificing accuracy of the extraction, may be shorter the time required for one extraction step.

Next, a second embodiment in which the extraction process is further shortened will be described using the wheel abnormality detection device of the first embodiment described above as a basic form. In this embodiment, the extraction means 3A of the first embodiment is replaced with an extraction means 3B. The extraction means 3B is continuously measured by the vibration measurement means 1 as shown in a block diagram in FIG. the vibration intensity data a ₁ ~a ₄ for each wheel, which is, by sampling at a predetermined fixed time interval [delta] _f, fixed to generate a measurement waveform X ₂₁ to X ₂₄ consisting of a data string arranged in time order For the sampling processing unit 21 and each wheel, only the component synchronized with the rotation of the wheel is extracted from the measurement waveforms X _{21 to} X ₂₄ as a data sequence arranged in time order, and this is extracted as the synchronization component waveform b _1. The adaptive digital filter unit 22 that outputs to the abnormality warning means 4 as ˜b ₄ (or Y ₂₁ ˜Y ₂₄ ) and the model waveform R ₂ that becomes the model of the extraction of each wheel are generated as a data sequence arranged in time order Model waveform raw It becomes more a part 23.

As is apparent from a comparison between the block diagram for the extracting unit 3A of the first embodiment shown in FIG. 3 and the block diagram for the extracting unit 3B of the present embodiment shown in FIG. The ADF (r) of the adaptive digital filter unit 12 in the first embodiment inputs a different model waveform R _1r for each wheel, whereas in the second embodiment, it is common to these wheels. One model waveform R ₂ is input and used to generate a synchronous component waveform Y _2r in each ADF (r). This further reduces the processing time required for extraction, as will be described later. It is something that is going to be realized.

As a requirement that the model waveform R ₂ for this purpose should have, the synchronous component waveform Y _2r cannot be extracted with high accuracy for all the wheels even if it is common to each wheel. In the following, the configuration of the model waveform R ₂ that can satisfy this requirement will be described.

FIG. 13 is a graph showing a part of the model waveform R ₂ generated by the model waveform generation unit 23 with the horizontal axis representing the data order along the time axis, and in FIG. 7, R ₂ (i−1), R ₂ (i) and R ₂ (i + 1) respectively indicate i−1, i, and i + 1th data that are continuously arranged in time order in the data string.

In this embodiment, each data R ₂ (i) of the model waveform R ₂ common to each wheel is a frequency domain waveform obtained by Fourier transforming a time domain waveform arranged at the fixed time interval δ _f. As shown in FIG. 14, for any wheel, it becomes a mountain at the position of the synchronization frequency of the order of extraction, and forms a valley between the synchronization frequencies of the adjacent orders, and becomes the mountain The top of the part extends as a point or flat surface in a frequency range of infinitesimal or finite width, and the height of the top has the same value for all the synchronization frequencies of the extraction. This point is the same as each model waveform R _1r (r = 1 to 4) used in the first embodiment, except that the model waveform R ₂ is different from these waveforms R _1r at the top. Extends as a flat surface with a finite width, not a point , And the this is a model waveform R _2, is because must function as a punching die which can be extracted synchronization frequency component for any wheel of different wheel speeds, therefore, the model waveform R ₂ The top of the nth order is extended to include the synchronization frequency f _r ⁽ⁿ⁾ for all wheels.

That is, of all the wheels, when the fundamental frequency (frequency is minimum) for the wheel at the highest vehicle speed is f _min and the fundamental frequency (frequency is maximum) for the wheel at the lowest vehicle speed is f _max , the nth order of the model waveform R ₂ It is necessary that the upper limit f _u ⁽ⁿ⁾ and the lower limit f _w ⁽ⁿ⁾ of the extension region at the top of the above satisfy the conditions of the expressions (2) and (3), respectively.

On the other hand, in order to generate a model waveform R ₂ that is a data sequence arranged in time order from the basic frequencies f ₁ ^{(1) to} f ₁ ⁽⁴⁾ calculated based on the wheel speed data v _{1 to} v ₄ , From these fundamental frequencies f ₁ ^{(1) to} f ₁ ⁽⁴⁾ , an extension region of the n-th apex is set based on the above equations (2) and (3), and an appropriate number of these setting regions are set. J, for example, approximating that it consists of a collection of 20 frequencies, and adding the sine waves for these frequencies to generate a time domain waveform R ₂ (t) as shown in equation (4), A time domain waveform may be sampled at a fixed sampling time δ _f to create discrete data, which is used as a model waveform. In Equation (5), fw ⁽¹⁾ and fu ⁽¹⁾ represent a lower limit frequency and an upper limit frequency for the extension region of the primary apex, and A ₂ is a predetermined constant.

As described above, the waveform can convert the model waveform R ₂ in the frequency domain, it has a top portion of finite width, extending region of the top portion of the plurality of wheels according to the extraction of the synchronizing component waveform, It can be determined based on both the fundamental frequency f _min of the fastest wheel and the fundamental frequency f _max of the _slowest wheel, but as a simpler means, instead of this, A is a predetermined numerical value. , F _min of 0 to + A%, or f _max of −A to 0%.

When set in this way, it is preferable because the extended frequency region of the top can be determined more easily based on only one wheel speed of f _max or f _min. If the vehicle is equipped with a vehicle that controls the wheel based on the wheel side of the wheel, the extension frequency region of the top is determined based only on the fundamental frequency f _min of the fastest wheel. By using the wheel number selected by a device other than the device, or the fundamental frequency f _min of the highest speed wheel collected by a device other than this device and the data correlated therewith, this device is extremely simple. Can be.

  Also, if there are characteristic values collected in real time for other purposes such as a characteristic representing the wheel speed instead of the wheel speed, such as a vehicle body yaw rate, these can be diverted.

  Here, the numerical value A may be set so that the n-th order synchronization frequency is located within the extension range of the top for any wheel under any driving condition of the vehicle. (%) Is a preferred range.

In order to confirm the success or failure of extraction of the component synchronized with the rotation of the wheel by the extraction means 3B configured as described above, sampling is performed at a fixed time interval δ _f from the vibration intensity data a ₁ of the traveling first wheel 90A. The measured measurement waveform X ₂₁ and the synchronization component waveform Y ₂₁ output from the digital filter unit, the data X ₂₁ (i) and Y ₂₁ (i) constituting these waveforms, and the time interval δ _A time domain waveform was generated by arranging them at _f , and these were further Fourier transformed to obtain a frequency domain waveform. The frequency domain waveform corresponding to the measurement waveform X ₂₁ is as shown in FIG. 9, and the frequency domain waveform corresponding to the synchronization component waveform Y ₂₁ is shown in FIG.

In addition, as a comparative example of the model waveform common to all wheels, the frequency domain waveform is the synchronous frequency of the second wheel with the second wheel as a representative wheel instead of the peak of each mountain in the frequency domain waveform being a flat surface. FIG. 16 shows a frequency domain waveform corresponding to the synchronous component waveform Y ₂₁ when extracted using the model waveform R ₂₂ having a point-like apex at the position.

Comparing FIG. 9, FIG. 15, and FIG. 16, FIG. 15 has peaks of height having the same value as the vibration intensity in FIG. 9 at the synchronization frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ . It has a waveform with a valley formed between them, and it shows that a synchronous component waveform consisting of the synchronous frequency of the wheel could be extracted from the measured waveform X ₁₁ for each wheel. If the frequency domain waveforms of the comparative example shown in FIG. 16, of course, the position of the fundamental frequency in the frequency domain waveform corresponding to the measured waveform X ₂₁ in FIG. 9, it and corresponding to the model waveform R ₂₂ used in the extraction As a result, the height of the peak, for example, the height of the peak at n = 1, is lower than that shown in FIG. 15 and the extraction of the synchronous component waveform is unsatisfactory.

The waveform in FIG. 9 is based on a sampling process in which the time interval δ _f at the time of sampling is 140 μs and the sampling number M is 2048, and the fundamental frequency at the time of measurement is 13.5 Hz.

Figure 17 is a flowchart illustrating a process of extracting means 3B of the second embodiment is executed, the extraction unit 3B as 4 wheel number, first, the model waveform generator 23 on the wheel speed data v ₁ ~ of the wheels v ₄ is input (step s201), and a model waveform R ₂ (i) (i = 1 to M, where M is the number of samplings used for one extraction) is generated based on the procedure described above. (Step s202).

Next, although the procedure for _obtaining the data of the synchronous component waveform Y _2r one by one in the order of data is performed, prior to this, the synchronous component waveform for each wheel is initialized (step s203). Next, after i = 1 (step s204), the first data R ₂ (i) of the model waveform R ₂ generated by the model waveform generation unit 23 is input to the ADF (1) (step s205).

Next, using this model waveform data R ₂ (1), the first data Y _2r (1) (r = 1 to 4) of the synchronous component waveform for all of the 1 to 4 wheels is calculated. as r = 1 to start it from the first wheel (step s206), the processing unit of the fixed sampling processing section 21 (1), enter the vibration intensity data a ₁ (step s 207), then this time interval δ Sampling at _f generates measurement waveform data X ₂₁ (i) (step s208), and then a difference E ₂₁ (1) is obtained (step s109), which is input to ADF (1).

In ADF (1), the filtering parameter is optimized by the difference E ₂₁ (1), and the first input of the synchronous component waveform Y ₂₁ from the previously input R ₂₁ (1) and the difference E ₂₁ (1). Data Y ₂₁ (1) is obtained (step s210), used to calculate the next difference E ₂₁ (2), and output to the abnormality warning device 4 (step s211).

This completes the round of steps for obtaining Y ₂₁ (1), and then executes the same steps for the second and subsequent wheels, so r = r + 1 (step s212) until r = 5 This is repeated (step s213). When the above processing for i = 1 is completed, i = i + 1 is set (step s214), and the above is repeated until i = M (step s215) to complete one extraction step.

As described above, when calculating the synchronizing component waveform Y _11, the vibration intensity data a _r entered, without performing the time-consuming processing such as Fourier transform, as it is sampled at predetermined time intervals [delta] _f, It is the same as the first embodiment in that it can be processed in the order of input to generate the synchronous component waveform Y _1r , and one extraction process can be completed in a short time. In the apparatus, the processing time can be made shorter than that of the first embodiment in the following points.

This is in step s205, is due to the model waveform R ₂ were of common to all the wheels, the first effect of this is, in the first embodiment, the extraction of the wheels of the synchronization component waveform since respectively using separate model waveform data R _1r in each time to calculate the respective synchronous component waveform data Y _1r (i) for each wheel, must call model waveform data R _1r and (i) from the memory or buffer In contrast, in this embodiment, R ₂ (i) can be called from the memory or buffer only once for all four wheels Y _2r (i) (r = 1 to 4). Can be a point. That is, with M as the number of samplings and R as the number of wheels, in the first embodiment, it was necessary to call the model waveform M · R times in total, whereas in the second embodiment, And you can do this M times.

The second effect obtained by making the model waveform R ₂ common to all the wheels is that the number of processes required for generating the model waveform R ₂ in step s202 can be reduced. First embodiment In the second embodiment, a separate model waveform R _1r had to be generated for each wheel. However, in the second embodiment, it is only necessary to generate a common model waveform R ₂ once. This also makes it possible to shorten the extraction process.

  For the other processing parts, the number of processing steps is the same, and accordingly, the entire processing time can be shortened by the amount of the processing number reduced in the first and second points.

Next, a third embodiment will be described below. The third embodiment has a configuration in which the extraction unit 3A of the first embodiment is replaced with an extraction unit 3C. As shown in a block diagram in FIG. The vibration intensity data a _{1 to} a ₄ measured for each wheel are sampled at a time interval δ _vr that changes according to the wheel speed, respectively, and measurement waveforms X _{31 to} X consisting of a data sequence arranged in time order are sampled. _For the variable sampling processing unit 31 that generates ₃₄ and each wheel, only the components synchronized with the rotation of the wheel are extracted from the measurement waveforms X _{31 to} X ₃₄ as a data sequence arranged in time order, An adaptive digital filter unit 32 that outputs the synchronous component waveforms b _{1 to} b ₄ (or Y _{31 to} Y ₃₄ ) to the abnormality warning means 4 and a model waveform R ₃ that is a model of the extraction of each wheel are arranged in time order. Generate as a data column The more the model waveform generator 33 that.

As is apparent from a comparison between the block diagram for the extraction unit 3A of the first embodiment shown in FIG. 3 and the block diagram for the extraction unit 3C of the present embodiment shown in FIG. In the first embodiment, the vibration waveform data a _r is sampled at a fixed time interval δ _f and the ADF corresponding to each wheel of the adaptive digital filter unit 12 inputs a different model waveform R _1r for each wheel. On the other hand, in the third embodiment, sampling is performed at a variable time interval δ _vr that changes with time according to the wheel speed. Instead, the ADF has a time-invariant, In addition, the configuration is such that a common model waveform R ₃ is input to each wheel.

The variable sampling processing unit 31 includes processing units (1) to (4) corresponding to the wheels 90A to 90D. For example, in the processing unit (1), the vibration intensity illustrated with time on the horizontal axis in FIG. The data waveform a ₁ is sampled at a variable time interval δ _v1 according to the wheel speed v ₁ that changes with time, and the sampled data is shown in a graph in FIG. 20, with the horizontal axis representing the data order along the time axis. as shown, it outputs a measured waveform X ₁₁ consisting of data string time-sequenced. In FIG. 5, X ₁₁ (i-1), X ₁₁ (i), and X ₁₁ (i + 1) are respectively arranged in a time sequence continuously to form a data string, i-1, i, i + The first data is shown.

The time interval δ _vr for the r-th wheel is a predetermined value that is a basic frequency f _r (1) corresponding to the cycle of one rotation of the wheel and the number of divisions when equally dividing the rotation of the wheel. Set as the inverse of the product of k. That is, it is set to sample every 1 / k rotation of the wheel, and this is the first time that the model function R ₃ is made invariant with time by setting the number of data in one wheel revolution to be constant. Because it can.

  The sampling number M for one extraction step uses a predetermined value, but includes 2 to 5 rotations of the wheel to make the extraction process using the adaptive digital filter have sufficient accuracy. At the same time, the division number k representing the sampling number for one round of the tire is preferably 360 to 3600, and therefore the sampling number M is preferably 720 to 18000.

In order to calculate the variable time interval δ _vr , for each wheel, a δ _v calculation unit (r) for calculating the time interval δ _vr from the input wheel speed v _r is a processing unit of the variable sampling processing unit 31 ( provided before r).

Adaptive digital filter unit 12 is made of an adaptive digital filter corresponding to the respective wheels ADF (1) ~ (4) , each ADF, for example, the first wheel of the ADF (1) is a model waveform R ₃ fixed type, the E ₃₁ the difference between the measured waveform X ₃₁ and synchronizing component waveform Y _31, and generates a synchronizing component waveform Y ₃₁ on the basis of the model waveform R ₃₁ input the difference E _31, synchronization component waveform Y ₃₁ Is the same as that described in the first embodiment in that it is configured to adaptively control the filtering parameter when generating the difference E ₃₁ , and detailed description thereof is omitted.

A predetermined fixed frequency is defined as a fixed fundamental frequency f ₀ , and a frequency obtained by multiplying the fixed fundamental frequency by a natural number n is referred to as an “nth-order fixed synchronization frequency”, and the fixed fundamental frequency f ₀ and the number of divisions When the reciprocal of the product of k is the time interval δ ₀ , the fixed model waveform R ₃ generated by the model waveform generation unit 33 is obtained by arranging the data of the model waveform R ₃ every time interval δ ₀ . The frequency domain waveform formed by Fourier transform of the time domain waveform is a peak at the position of the fixed synchronization frequency f ₀ ⁽ⁿ⁾ = n · f ₀ of the extraction order n, and between the fixed synchronization frequencies of the adjacent orders. In the shape of a valley, the peak of the peak extends as a point or flat surface in a frequency range of infinitesimal or finite width, and the height of the peak is fixed for all orders related to the extraction. Has the same value for the synchronization frequency As constructed.

That is, in FIG. 21, the frequency waveform corresponding to the model waveform R ₃ has frequencies f ₀ ^{(1) to} f ₀ ⁽²⁰⁾ (where f is an integer multiple n (n = 1 to 20) of the fixed fundamental frequency f ₀ ^). ₀ ⁽ⁿ⁾ = n · f ₀ ⁽¹⁾ ), and has a shape that becomes a valley between these peaks. In the case shown in FIG. 8, the top of the portion that becomes the peak is infinitely small. It is arranged as a point, and the height of the top portion has a shape having the same value at all frequencies of f ₀ ^{(1) to} f ₀ ⁽²⁰⁾ .

On the other hand, in order to generate the model waveform R ₃ that is a data sequence arranged in time order from the fixed fundamental frequency f ₀ , for example, the time domain waveform R ₃ (t) is calculated by the equation (6). The waveform may be sampled at a time interval δ _f to form a data string along the time axis. Here, N represents the maximum order related to the extraction, and A ₃ is a predetermined constant.

In order to confirm the success or failure of extraction of the component synchronized with the rotation of the wheel by the extraction means 3C configured as described above, the measurement sampled at the variable time interval δ _v1 from the vibration intensity data a ₁ of the traveling wheel 90A. Each of the waveform X ₃₁ and the synchronous component waveform Y ₃₁ output from the digital filter unit is replaced with each data X ₃₁ (i) and Y ₃₁ (i) constituting these waveforms at a fixed time interval δ _0. Were arranged side by side to generate a time domain waveform, and these were further Fourier transformed to obtain a frequency domain waveform. Figure 22 is a frequency domain waveform corresponding to the measured waveform X _31, FIG. 23 is a graph showing a frequency domain waveform corresponding to the synchronizing component waveform Y _31.

FIG. 23 is compared with FIG. 23. FIG. 23 is composed of peaks having a height having the same value as the vibration intensity in FIG. 22 at the synchronization frequencies f ₀ ^{(1) to} f ₀ ⁽²⁰⁾ . Shows a waveform with a valley formed, and shows that the waveform synchronized with the rotation of the wheel, that is, the waveform consisting of the synchronization frequency of the wheel was extracted from the measured waveform X _31. In the abnormality warning device 4, this height is set to a predetermined limit by performing processing for determining the height of the peaks at the synchronization frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ or processing equivalent thereto. When the value is exceeded, it can be determined that some abnormality has occurred in the wheel.

Incidentally, used in the above calculation, the time interval [delta] _v1 in the sampling of the vibration intensity data a _1, a 118Myuesu, fixed fundamental frequency f ₀ is 16.5 Hz, the division number k was 512.

FIG. 24 is a flowchart showing the processing executed by the extraction means 3C of the third embodiment. The extraction means 3 first sets the wheel number r to 1 with the number of wheels set to 4 (step s301). For the wheels, wheel speed data v ₁ is input (step s102), then the basic frequency of the first wheel is obtained based on the wheel speed data v ₁ to calculate a variable time interval δ _v1 in sampling (step s303), Thereafter, the synchronous component waveform is initialized (step s304).

Then, calculate the i-th data Y ₃₁ synchronization component waveform Y ₃₁ from the i-th data X ₃₁ of the i-th data R ₃ of model waveform R ₃ (i) and measuring the waveform X ₃₁ (i) (i) This follows the following procedure. That is, after initializing the data number and setting i = 1 (step s305), the model waveform data R ₃ (1) is input to the ADF (1) (step s306). While taking the intensity data a ₁ (step s307), this is sampled at the time interval δ _v1 to generate the first data X ₃₁ (1) of the measurement waveform X ₃₁ (step s308), and the difference E ₃₁ (1 ) (Step s309) and this is input to ADF (1).

In ADF (1), the filtering parameter is optimized by the difference E ₃₁ (1), and the first input of the synchronous component waveform Y ₃₁ from the previously input R ₃ (1) and the difference E ₃₁ (1). Data Y ₃₁ (1) is obtained (step s310), and this is output and used to calculate the next difference E ₃₁ (2) and is output to the abnormality warning device 4 (step s311).

With the above, one round of steps for obtaining Y ₃₁ (1) is completed, i = i + 1 is set (step s312), and Y ₁₁ (2) is entered, but before that, i is the sampling number M. determines whether the reached (step s313), if they reached if completes the extraction of the sync component waveform Y ₃₁ for the first wheel, as r = r + 1 (step s314), the second wheel for, was extracted synchronization component waveform Y ₃₂ in the same procedure, this is repeated until the last wheel (step s315), to complete the single extraction step. During the traveling, this extraction process is repeated at a predetermined cycle.

  In addition, the processing for each wheel is assumed to be serially connected in time for convenience of explanation, but can be parallel processing as long as the processing time of the apparatus is allowed. Can be shortened.

As described above, as in the first embodiment, when the synchronous component waveform Y _3r is extracted, the input vibration intensity data a _r is not subjected to time-consuming processing such as Fourier transformation, and the time interval δ _vr Can be processed in the order of input to generate the synchronized component waveform Y _3r , and the extraction process can be shortened without sacrificing the accuracy in extracting the components synchronized with the wheels. .

Next, a fourth embodiment in which the extraction process is further shortened will be described based on the wheel abnormality detection device of the third embodiment described above as a basic form. In this embodiment, the extraction means 3C of the third embodiment is replaced with an extraction means 3D. The extraction means 3D is continuously measured by the vibration measurement means 1 as shown in a block diagram in FIG. Each of the measured vibration intensity data a _{1 to} a ₄ for each wheel is sampled at a variable time interval δ _va that changes with time according to the wheel speed, and a measurement waveform X ₄₁ comprising a data sequence arranged in time order For the variable sampling processing unit 41 that generates ~ X ₄₄ and for each wheel, only the components synchronized with the rotation of the wheel are extracted from the measurement waveforms X _{41 to} X ₄₄ as a data sequence arranged in time order. Are output to the abnormality warning means 4 as synchronization component waveforms b _{1 to} b ₄ (or Y _{41 to} Y ₄₄ ), and a model waveform R _{4 serving} as a model for the extraction of each wheel is converted into time. De in order The more the model waveform generating unit 43 for generating a data sequence.

As is apparent from a comparison between the block diagram for the extracting unit 3A of the third embodiment shown in FIG. 18 and the block diagram for the extracting unit 3D of the fourth embodiment shown in FIG. The sampling time interval δ _vr in the variable sampling processing unit 31 in the third embodiment differs for each wheel, whereas the sampling time interval δ _a in the fourth embodiment uses a common value for all wheels. that is has different, by (of course, [delta] _a change with time) this, those of the fourth embodiment, as will be described in greater detail below, to further shorten the time required for the extraction process be able to.

The variable sampling processing unit 31 includes processing units (1) to (4) corresponding to the wheels 90A to 90D. For example, in the processing unit (1), the vibration shown in FIG. The intensity data waveform a ₁ is sampled at a variable time interval δ _va according to the wheel speed v _a that changes with time, and the sampled data is shown in FIG. 27, with the horizontal axis representing the data order along the time axis. As shown, the measurement waveform X ₄₁ composed of a data sequence arranged in time order is output. In FIG. 27, X ₄₁ (i−1), X ₄₁ (i), and X ₄₁ (i + 1) are respectively arranged in a time sequence continuously to form a data string, i−1, i, i + The first data is shown.

The wheel speed v _{a that} is the basis for calculating the time interval δ _{va ,} or the fundamental frequency f _a corresponding thereto, changes according to the wheel speeds of other wheels or other wheel speeds. For example, this may be calculated from the speed of the car, or may be a fundamental frequency with respect to the average value of the wheel speeds of all the wheels. Then, the time interval [delta] _va at this time is determined as the reciprocal of the product of the fundamental frequency f _a which is set this way, the predetermined value k as the number of divisions when equally dividing the wheel 1 rotating It is done.

Here, a predetermined fixed frequency is defined as a fixed fundamental frequency f ₀ , and a frequency obtained by multiplying the fixed fundamental frequency by a natural number n is referred to as an “nth-order fixed synchronization frequency”, and the fixed fundamental frequency f ₀ and the division Let the reciprocal of the product of the number k be the time interval δ ₀ .

FIG. 28 shows a frequency obtained by Fourier-transforming a time domain waveform in which each data R ₄ (i) of the model waveform R ₄ used for extraction of the synchronous component waveform Y _4r is arranged every predetermined time interval δ _0. This frequency domain waveform has a peak at the position of the fixed synchronization frequency f ₀ ⁽ⁿ⁾ = n · f ₀ of the order n related to the extraction, and a trough between the fixed synchronization frequencies of the adjacent orders. The top of the peak portion extends as a point or flat surface in a frequency range of infinitesimal or finite width, and the height of the top corresponds to the fixed synchronization frequency of all orders related to the extraction. In contrast, this is the same as the model waveform R ₃ used in the third embodiment.

However, that model waveform R ₄ is different from the waveform R ₃ is not a top a point, a point extending as a flat surface having a finite width, the model waveform R _4, compared which wheels of different wheel speeds This is because it is necessary to satisfy the requirements for functioning as a punching die that can extract the synchronization frequency component.

That is, among all the wheels, such a requirement is satisfied when the basic frequency (frequency is minimum) for the wheel at the highest vehicle speed is f _min and the basic frequency (frequency is maximum) for the wheel at the lowest vehicle speed is f _max. model waveform upper limit of the extension region of the n-th top of the R ₄ f _u ⁽ⁿ⁾ and the lower limit f _w ^(n), respectively, it is necessary to satisfy the condition of formula (2) and (3).

On the other hand, in order to generate the model waveform R ₄ so that the frequency domain waveform satisfies the above-described requirements, a known f _o , f _min , f _max is generated based on the equations (7) and (8). , F _a , an extension region of the top of the n-th order is set, and this setting region is approximated to an appropriate number J, for example, a set of 20 frequencies, and sine waves for these frequencies are added. In addition, a time domain waveform R ₄ (t) shown in Equation (9) is generated, and this time domain waveform is sampled at the fixed time interval δ ₀ to create discrete data, which is used as a model waveform. In Equation (5), fw ⁽¹⁾ and fu ⁽¹⁾ represent a lower limit and an upper limit for the extension region of the primary apex, and A ₄ is a predetermined constant.

The extension region at the top of the waveform formed by converting the model waveform R ₄ into the frequency domain is similar to that described in the second embodiment, and the basic frequency f _min of the highest speed wheel and the lowest speed wheel. It can be determined based on both the fundamental frequency f _max and, as a simpler means, instead of A as a constant, 0 to + A% of f _min or -A to 0% of f _max Furthermore, characteristics correlating with the wheel speed such as the vehicle body yaw rate can also be used.

The configured extractor 3D as described above, to confirm the success or failure of extraction of components synchronizing the rotation of the wheel, the vibration intensity data a ₁ of the first wheel 90A during running, corresponding to the second wheel 90B With respect to the measurement waveform X ₄₁ generated by sampling at the time interval δ _va based on the fundamental frequency f2 (1) and the synchronous component waveform Y ₄₁ output from the digital filter unit 42, each data X ₄₁ ( A time domain waveform is generated by arranging i) and Y ₄₁ (i) at a time interval δ ₀ , and further, a frequency domain waveform is calculated by Fourier transforming these time domain waveforms. Figure 29 shows the frequency domain waveform corresponding to the measured waveform X _41, Fig. 30 shows a frequency domain waveform corresponding to the synchronizing component waveform Y _21.

When FIG. 29 is compared with FIG. 30, FIG. 30 has a height crest having the same value as the vibration intensity in FIG. 29 at the synchronization frequencies f ₁ ^{(1) to} f ₁ ⁽²⁰⁾ . and forms a waveform in the form of troughs are formed on, in to generate a measurement waveform X ₄₁ of the first wheel 90A, and sampled using a time interval based on a wheel speed of the second wheel 90B Nevertheless, it shows that the expected extraction was performed.

The time interval δ _v2 in the sampling of the vibration intensity data a ₁ used for the above calculation is
The fixed fundamental frequency f ₀ was 16.5 Hz and the division number k was 512.

FIG. 11 is a flowchart showing the flow of processing executed by the extraction means 3D of the fourth embodiment. The extraction means 3D first sets the number of wheels to 4, and first sends the wheel speed data v of each wheel to the model waveform generation unit 43. _{1 to} v ₄ are input (step s401), and the model waveform R ₄ (i) (i = 1 to M, where M is the number of samplings used for one extraction) is input in accordance with the procedure described above based on this. Generate (step s402).

Then, although the procedure for _obtaining the data of the synchronous component waveform Y _4r one by one in the order of data is performed, prior to this, the synchronous component waveform for each wheel is initialized (step s403) and the wheel speeds v _{1 to} v _{Based on 4} (or this substitute characteristic), the variable sampling time interval δ _va is calculated (step s404). Next, after setting i = 1 (step s405), the first data R ₄ (i) of the model waveform R ₄ generated by the model waveform generation unit 43 is input to the ADF (1) (step s406).

Next, using this model waveform data R ₄ (1), the first data Y _4r (1) (r = 1 to 4) of the synchronous component waveform for all of the 1 to 4 wheels is calculated. as r = 1 to start it from the first wheel (step s407), the processing unit of the fixed sampling processing section 41 (1), enter the vibration intensity data a ₁ (step s408), then this time interval δ Sampling with _va generates measurement waveform data X ₄₁ (i) (step s409), and obtains a difference E ₄₁ (1) (step s410), which is input to ADF (1).

In ADF (1), the filtering parameter is optimized by the difference E ₄₁ (1), and the first input of the synchronous component waveform Y ₄₁ from the previously input R ₄₁ (1) and the difference E ₄₁ (1). Data Y ₄₁ (1) is obtained (step s411), is used to calculate the next difference E ₄₁ (2), and is output to the abnormality warning device 4 (step s412).

In this way, the round of steps for obtaining Y ₄₁ (1) is completed, and then the same steps are performed for the second and subsequent wheels, so r = r + 1 (step s413) until r = 5 This is repeated (step s414). When the above processing for i = 1 is completed, i = i + 1 is set (step s415), and this is repeated until i = M (step s416), thereby completing one extraction step.

As described above, when calculating the synchronous component waveform Y _4r , the input vibration intensity data a _r is sampled at a predetermined time interval δ _va without performing time-consuming processing such as Fourier transform, It is the same as the third embodiment in that it can be processed in the input order to generate the synchronous component waveform Y _4r and the extraction process can be completed in a short time, but in the apparatus of the fourth embodiment, In the following points, the processing time can be made shorter than that of the third embodiment.

This is due to the use of the time interval δ _va common to each wheel when generating the measurement waveform data X _4r (i) in step s409. The first effect of this is the third embodiment. Since the sampling time interval δ _vr differs depending on the wheel, the i-th synchronization component waveform data Y4r (i) cannot be synchronized between the wheels to obtain the i-th synchronization component waveform data Y4r (i). For this reason, the model waveform data R _1r (i) must be called from the memory or buffer each time the synchronous component waveform data Y ₄₁ (i) is calculated for each wheel. In this embodiment, the processing of the i-th data of all the wheels can be completed by calling R ₄ (i) from the memory or buffer only once. That is, with M as the number of samplings and R as the number of wheels, in the third embodiment, it was necessary to call the M / R model waveform in all, whereas in the fourth embodiment Can do this M times.

The second effect of having the time interval δ _va common to all the wheels is that the number of processes required to generate the time interval δ _va in step s404 can be reduced. In the third embodiment, Since a separate time interval δ _vr had to be generated for each wheel, in the fourth embodiment, it is only necessary to calculate one common time interval δ _va , Can also contribute to shortening the extraction process.

  The wheel abnormality detection device of the present invention can be used to detect wheel abnormality in various vehicles.

It is the plane schematic diagram which shows arrangement | positioning of a wheel abnormality detection apparatus, and the side surface schematic diagram which shows the abnormal part which arose in a part of wheel. It is a block diagram which shows the structure of a wheel abnormality detection apparatus. It is a block diagram which shows the extraction means of the vehicle abnormality detection apparatus of 1st embodiment. It is a graph which shows vibration intensity data waveform a1 concerning a _first embodiment. It is a graph showing a portion of a measurement waveform X ₁₁ according to the first embodiment. It is a graph showing a portion of a synchronous component waveform Y ₁₁ according to the first embodiment. It is a graph showing a portion of the model waveform R ₁₁ according to the first embodiment. It is a graph showing a frequency domain waveform corresponding to the model waveform R ₁₁ according to the first embodiment. It is a graph showing a frequency waveform corresponding to the measured waveform X ₁₁ according to the first embodiment. It is a graph showing a frequency waveform corresponding to the synchronizing component waveform Y ₁₁ according to the first embodiment. It is a flowchart which shows the process of the extraction means which concerns on 1st embodiment. It is a block diagram which shows the extraction means of the vehicle abnormality detection apparatus of 2nd embodiment. It is a graph showing a portion of the model waveform R ₂₁ according to the second embodiment. It is a graph showing a frequency domain waveform corresponding to the model waveform R ₂₁ according to the second embodiment. It is a graph showing a frequency waveform corresponding to the measured waveform X ₂₁ according to the second embodiment. It is a graph showing a frequency waveform corresponding to the second according to the embodiment synchronizing component waveform Y _21. It is a flowchart which shows the process of the extraction means which concerns on 2nd embodiment. It is a block diagram which shows the extraction means of the vehicle abnormality detection apparatus of 3rd embodiment. Is a graph showing the vibration intensity data waveform a ₁ according to the third embodiment. It is a graph showing a portion of a measurement waveform X ₃₁ according to the third embodiment. It is a graph showing a frequency domain waveform corresponding to the model waveform R ₃₁ according to the third embodiment. It is a graph showing a frequency waveform corresponding to the measured waveform X ₃₁ according to the third embodiment. It is a graph showing a frequency waveform corresponding to the synchronizing component waveform Y ₃₁ according to the third embodiment. It is a flowchart which shows the process of the extraction means which concerns on 3rd embodiment. It is a block diagram which shows the extraction means of the vehicle abnormality detection apparatus of 4th embodiment. Is a graph showing the vibration intensity data waveform a ₁ according to the fourth embodiment. It is a graph showing a portion of a measurement waveform X ₄₁ according to the fourth embodiment. It is a graph showing a frequency domain waveform corresponding to the model waveform R ₄₁ according to the fourth embodiment. It is a graph showing a frequency waveform corresponding to the measured waveform X ₄₁ according to the fourth embodiment. It is a graph showing a frequency waveform corresponding to the synchronizing component waveform Y ₄₁ according to the fourth embodiment. It is a flowchart which shows the process of the extraction means which concerns on 4th embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Vibration measuring means 1a-1d Vibration sensor 2 Wheel speed measuring means 2a-2d Wheel speed sensor 3, 3A, 3B, 3C, 3D Extraction means 4 Abnormal warning means 4a Abnormal control part 4b Abnormal display part 5 Central controller 10 Wheel abnormality Detection apparatus 11 Fixed sampling processing unit 12 Adaptive digital filter unit 13 Model waveform generation unit 21 Fixed sampling processing unit 22 Adaptive digital filter unit 23 Model waveform generation unit 31 Variable sampling processing unit 32 Adaptive digital filter unit 33 Model waveform generation unit 41 Variable sampling Processing unit 42 Adaptive digital filter unit 43 Model waveform generation unit 90a to 90d Wheel 91 Delamination

Claims (10)

For a plurality of wheels mounted on a running vehicle, a vibration measuring means for continuously measuring the vibration intensity of each wheel, and a vehicle that correlates the rotational speed of each wheel directly or with the wheel speed. Wheel speed measuring means for measuring continuously and indirectly by measuring characteristics; and extracting means for extracting a waveform synchronized with the rotation of the wheel from the data measured by the vibration measuring means and the wheel speed measuring means; In the wheel abnormality detection device comprising the abnormality warning means for determining the state of the wheel from the waveform extracted by the extraction means and warning the abnormality when this state is outside the predetermined normal range,
The extraction means includes
Fixed to generate a “measurement waveform” consisting of a data sequence arranged in time order by sampling vibration intensity data for each wheel continuously measured by the vibration measuring means at a predetermined fixed time interval δ _f A sampling processing unit;
For each wheel, only the component synchronized with the rotation of the wheel is extracted from the measurement waveform as a data sequence arranged in time order, and this is output as the “synchronous component waveform” to the abnormality warning means. A filter section;
It consists of a model waveform generation unit that generates a “model waveform” as a model of the extraction of each wheel as a data sequence arranged in time order,
The adaptive digital filter unit receives the model waveform, generates the synchronization component waveform based on both the difference between the measurement waveform and the synchronization component waveform, and the input model waveform, and the synchronization component waveform Is configured to adaptively control the filtering parameter in generating the difference using the difference,
When the frequency obtained by multiplying the fundamental frequency corresponding to the period of one rotation of each wheel by a natural number n is the `` nth synchronization frequency '' for that wheel,
The model waveform for each wheel, frequency domain waveform of time domain waveform obtained by arranging the data in the time interval of the [delta] _f can be the Fourier transform of this model waveform, the wheel, the next number of the synchronization frequency according to the extracted It becomes a mountain at the position of, and forms a valley between adjacent synchronization frequencies of the order, the top of the portion that becomes the mountain extends to a frequency range of infinitesimal or finite width as a point or flat surface, The height of the top is configured to have the same value for all orders of synchronization frequency for the extraction,
The model waveform generation unit sets a basic frequency of each wheel or a range including the basic frequency based on data from the wheel speed measuring unit, and sets the frequency domain waveform characteristics from the set basic frequency or the range. A wheel abnormality detection device configured to generate a model waveform that satisfies the above. The adaptive digital filter unit is configured to generate respective synchronous component waveforms for a plurality of wheels based on a single model waveform common to them,
In the frequency domain waveform corresponding to the common model waveform, the frequency range of the peak corresponding to the n-th synchronization frequency is the n-th order for all wheels related to the extraction of the synchronization component waveform based on the common model waveform. The wheel abnormality detection device according to claim 1, further comprising:   Assuming that A is a predetermined percentage, the frequency range at the top of the peak corresponding to the n-th synchronization frequency is the basic frequency among the plurality of wheels related to the extraction of the synchronization component waveform based on the common model waveform. The wheel abnormality detection device according to claim 2, wherein the wheel abnormality detection device is set as 0 to + A% of the n-th order synchronization frequency of the minimum wheel.   The frequency range at the top of the mountain corresponding to the n-th synchronization frequency is a wheel having a minimum fundamental frequency among a plurality of wheels related to extraction of a synchronization component waveform based on the common model waveform, and the fundamental frequency is The wheel abnormality detection device according to claim 2, wherein the wheel abnormality detection device is set based on the n-th synchronization frequency for both wheels with the largest wheel.   Each time the adaptive digital filter unit inputs data constituting the common model waveform, for each of all wheels related to extraction of the synchronous component waveform based on the common model waveform, The wheel abnormality detection device according to any one of claims 2 to 4, wherein the wheel abnormality detection device is configured to perform a process of outputting only one data. For a plurality of wheels mounted on a running vehicle, a vibration measuring means for continuously measuring the vibration intensity of each wheel, and a vehicle that correlates the rotational speed of each wheel directly or with the wheel speed. Wheel speed measuring means for measuring continuously and indirectly by measuring characteristics; and extracting means for extracting a waveform synchronized with the rotation of the wheel from the data measured by the vibration measuring means and the wheel speed measuring means; In the wheel abnormality detection device comprising the abnormality warning means for determining the state of the wheel from the waveform extracted by the extraction means and warning the abnormality when this state is outside the predetermined normal range,
The extraction means includes
For each wheel, the vibration intensity data continuously measured along the time axis by the vibration measuring means is sampled at a time interval δ _v that changes according to the data from the wheel speed measuring means, and the data is arranged in time order. A variable sampling processing unit that generates a "measurement waveform" consisting of columns;
For each wheel, only the component synchronized with the rotation of the wheel is extracted from the measurement waveform as a data sequence arranged in time order, and this is output as the “synchronous component waveform” to the abnormality warning means. A filter section;
It consists of a model waveform generation unit that generates a “model waveform” as a model of the extraction of each wheel as a data sequence arranged in time order,
The time interval δ _v used to generate the measurement waveform for each wheel in the variable sampling processing unit corresponds to a cycle of one rotation of the wheel calculated based on data from the wheel speed measuring means. Or a fundamental frequency f _a corresponding to a cycle of one rotation of the virtual wheel calculated by a predetermined procedure based on data from the wheel speed measuring means, and one rotation of these wheels. It is obtained as the reciprocal of the product with a predetermined value k which is the number of divisions when equally dividing,
The adaptive digital filter unit receives the model waveform, generates the synchronization component waveform based on both the difference between the measurement waveform and the synchronization component waveform, and the input model waveform, and the synchronization component waveform The filtering parameter when generating is adaptively controlled using the difference, and each synchronous component waveform for a plurality of wheels is configured to be generated based on one model waveform common to these wheels,
A predetermined fixed frequency is set as a fixed basic frequency f ₀ , and a frequency obtained by multiplying the fixed basic frequency by a natural number n is referred to as an “n-th fixed synchronization frequency”, and the fixed basic frequency f ₀ and the division number k are When the reciprocal of the product of the time interval δ ₀ ,
The common model waveform is a frequency domain waveform obtained by performing Fourier transform on a time domain waveform in which each data of the model waveform is arranged at every time interval δ ₀ , and is a peak at the position of the fixed synchronization frequency of the order related to the extraction. Between the fixed synchronization frequencies of adjacent orders, forming a valley, and the top of the peak is extended as a point or flat surface in a frequency range of infinitesimal or finite width, and the height of the top The wheel abnormality detection device is configured to have the same value with respect to the fixed synchronization frequencies of all orders related to the extraction. The time interval δ _v used for the sampling process corresponding to each wheel is the same for all of the plurality of wheels related to the extraction of the synchronous component waveform, and the minimum of the fundamental frequencies of these wheels is used. F _min , the largest one is f _max , and the one used to calculate the time interval δ _v is f _a, and the frequency domain waveform corresponding to the model waveform corresponds to the nth-order fixed synchronization frequency The frequency range of the top of the peak is configured to include at least a range not less than n times f ₀ · (f _min / f _a ) and not more than n times f ₀ · (f _max / f _a ) The wheel abnormality detection device according to claim 6. Of fundamental frequency for all of the wheels according to the extraction of the synchronizing component waveform based on said common model waveform, the smallest ones, said, with use in the calculation of the common time interval [delta] _v, as a percentage which is determined a A pre The wheel abnormality detection device according to claim 7, wherein a frequency range of the peak corresponding to the n-th fixed synchronization frequency is set as a range of 0 to + A% of the n-th fixed synchronization frequency.   The frequency range at the top of the mountain corresponding to the n-th fixed synchronization frequency is set to the wheel having the lowest fundamental frequency and the largest to the fundamental frequency among the plurality of wheels related to the extraction of the synchronization component waveform based on the common model waveform. The wheel abnormality detection device according to claim 7, wherein the wheel abnormality detection device is set on the basis of both n-order synchronization frequencies with the wheel.   Each time the adaptive digital filter unit inputs data constituting the common model waveform, the data of the synchronous component waveform for each of all the wheels related to the extraction of the synchronous component waveform based on the common model waveform. The wheel abnormality detection device according to claim 7, wherein the wheel abnormality detection device is configured to perform a process of outputting only one of the two.

Images