JPH11271181A - Method and apparatus for diagnosing abnormalities in rolling bearings - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide a method and an apparatus for making it possible to diagnose an abnormality of a rolling bearing from an initial stage by using a sound pressure signal detected by a microphone even when specifications are unknown. SOLUTION: A sound generated from a rolling bearing is detected as a sound pressure by a microphone, only a sound pressure signal in a high frequency band including an ultrasonic band is extracted, an envelope signal is generated, and a spectrum distribution of the envelope signal is generated. Is calculated to calculate the spectrum level, and a predetermined judgment reference value is compared,
If the calculated spectral level is large, the rolling bearing is determined to be abnormal, and for the rolling bearing determined to be abnormal, the spectrum pattern of the envelope signal is calculated. If the spectral pattern is smooth, it is determined to be lubrication failure. If the spectrum pattern is steep, it is determined that a defect has occurred.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for diagnosing the presence or absence and the type of abnormality of a rolling bearing, which is a component of a rotary machine, using sound.

[0002]

2. Description of the Related Art Abnormality diagnosis of rolling bearings (hereinafter abbreviated as "bearings"), which is an important component of rotating machines such as blowers, pumps, and belt conveyor driving devices, is currently performed for reasons of high diagnostic accuracy. A method and an apparatus using vibration as a measurement parameter are mainly used. However, practical drawbacks have been pointed out, such as the high price of the vibration sensor and the high cost of wiring the signal cable. In addition, when performing vibration measurement using a hand-held vibration sensor, it is necessary to bring the sensor into contact with equipment, and thus there is a problem in terms of safety and workability. To compensate for these drawbacks, that is, diagnostic accuracy is inferior to vibration, but for the purpose of providing a cheaper, safer and more efficient diagnostic method and device than vibration, abnormality diagnosis using the sound generated from rolling bearings Can also be performed.

Conventionally, a method of diagnosing a bearing abnormality using sound has been performed, and is disclosed, for example, in Japanese Patent Application Laid-Open No. Sho 62-5137. In this method, the sound generated from the bearing is detected by a microphone, a signal in a frequency band of the natural frequency of the bearing (usually 1 kHz to 5 kHz) is extracted, and the quality of the bearing is determined from the signal level of this signal. .

[0004] Further, although the diagnosis target is not specified, as an apparatus for diagnosing a machine failure by sound, for example, Japanese Patent Application Laid-Open No. H2-22-2
No. 49927 is disclosed. In this device, an audible sound generated by the diagnosis target machine is detected by an acoustic sensor, and two predetermined specific frequency components are extracted.
The presence or absence of a failure or its degree is diagnosed based on the difference between the magnitudes of the two signal components.

[0005]

However, Japanese Patent Application Laid-Open No.
In the method disclosed in Japanese Patent Application Laid-Open No. 2-5137, diagnosis is to be performed using signals in the natural frequency band of the bearing (usually 1 kHz to 5 kHz). It is difficult to extract only the sound generated from the bearing because the frequency is also included. Therefore, there is a drawback that the accuracy of diagnosis cannot be ensured, and diagnosis cannot be performed unless it is at the end of abnormalities. This was a big handicap compared with the diagnosis using vibration as a measurement parameter.

Also, in the apparatus disclosed in Japanese Patent Application Laid-Open No. 2-249927, the frequency band that can be caught most sensitively as audible sound is 1 kHz to 4 kHz as described later. Includes similar issues. Furthermore, in order to determine the specific frequency to be extracted, it is necessary to know in advance the exact specifications of the machine to be diagnosed, for example, the number of revolutions of the machine and the bearing number of the bearing, and it is not possible to diagnose a machine with unknown specifications There was a disadvantage.

It is an object of the present invention to be able to diagnose a bearing abnormality at an early stage by using a sound pressure signal detected by a generally used microphone even when the specifications such as the number of revolutions and the identification number are unknown. The purpose of the present invention is to provide a method and an apparatus for making it easier.

[0008]

An abnormality diagnosis method according to the present invention detects a sound generated from a rolling bearing as a sound pressure by a microphone and extracts only a sound pressure signal in a high frequency band including an ultrasonic band. The envelope signal is generated, the spectrum distribution of the envelope signal is obtained, the spectrum level is calculated, and the spectrum level is compared with a predetermined determination reference value. If the calculated spectrum level is large, the rolling bearing is determined to be abnormal. In addition, for the rolling bearing determined to be abnormal as necessary, the spectrum pattern of the envelope signal is calculated.If the spectrum pattern is smooth, it is determined to be poor lubrication, and if the spectrum pattern is steep, It is characterized by determining that a defect has occurred.

The abnormality diagnosis apparatus according to the present invention includes a microphone for detecting a sound pressure, a high-frequency band-pass filter for extracting only a high-frequency band component of the detected sound pressure signal, and a sound pressure signal after passing the filter. An envelope processing unit that performs envelope processing, a spectrum analysis unit that determines the spectrum distribution of the envelope signal, a spectrum level calculation unit that determines the magnitude of the spectrum distribution, and compares the calculated spectrum level with a reference value to determine whether an abnormality has occurred. An abnormality presence / absence determination unit that determines presence / absence;
It is characterized by comprising a spectrum pattern calculation unit for calculating a spectrum pattern and an abnormality type determination unit for determining the type of abnormality.

It is to be noted that the high-frequency band in the method of the present invention should be a band of 10 kHz or more, and
As the high-frequency band-pass filter in the device of the present invention,
It is desirable to provide a filter having a cutoff frequency of 10 kHz. Further, as the envelope signal generating method in the method of the present invention, the absolute value processing of the sound pressure signal is performed, the noise is removed using an adaptive filter after passing through a low frequency band-pass filter, and the method of the present invention is performed. The envelope processing unit may be composed of an absolute value processing unit, a low-frequency band-pass filter, and an adaptive filter for removing noise. Further, as the spectrum pattern in the method of the present invention, an impact evaluation index of a spectrum waveform is adopted, and the spectrum pattern calculation unit in the apparatus of the present invention includes a mechanism for calculating an impact evaluation index of the spectrum waveform. Preferably,

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION A sound level meter (JIS C150) generally used to measure the sound generated from a bearing when the bearing is normal and when the bearing is abnormal (when the outer raceway is damaged).
2 or JIS C 1505) Detects the sound pressure as a sound pressure, performs 1/3 octave analysis of the sound pressure signal to obtain a frequency distribution image in the audible sound band, and applies the A-weighted filter to the result. As shown in FIG.

When diagnosing with an audible sound, the noise level (52 to 52) at the frequency of the peak spectrum (1 kHz in FIG. 2) is used.
Since the frequency component lower than 53 dB (A) by 10 dB or more is hardly audible to the ear, the actually audible sound is a sound of about 1 kHz to 4 kHz in the example of FIG. The sounds are practically almost indistinguishable.

FIG. 3 shows an example in which a sound pressure signal is directly Fourier transformed to obtain a sound pressure spectrum distribution. The frequency band in which there is a clear difference in sound pressure between the normal state and the abnormal state is a band of 10 kHz or more, a high frequency band including the so-called ultrasonic band, and the sound pressure signal in this band determines whether the bearing is normal or abnormal. Diagnosis can be made. As described later, there is no characteristic peak spectrum in the spectrum of the sound pressure envelope signal of the poorly lubricated bearing. That is, the spectrum distribution becomes like white noise. In contrast,
A periodic peak spectrum is generated in the spectrum of the sound pressure envelope signal of the damaged bearing. Therefore, even if the specifications of the bearing are unknown, the type of abnormality such as poor lubrication and defects such as scratches can be determined.

In the present invention, the diagnosis is specifically performed by the following method. After the sound generated from the bearing is detected and amplified by a microphone as a sound pressure, a sound pressure signal in only a high frequency band including an ultrasonic band is extracted to create an envelope signal thereof. Then, the spectrum distribution of the envelope signal is obtained. A judgment reference value corresponding to the spectrum level when the bearing is normal is stored in advance, and the spectrum level at the time of diagnosis is compared with this to determine that the bearing is abnormal if it is larger than the judgment reference value. Further, the spectrum pattern of the envelope signal is calculated for the bearing determined to be abnormal, and if the pattern is smooth, the type of abnormality is determined to be lubrication failure, and if the pattern is steep, it is determined to be defective.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail based on embodiments. FIG. 1 is a block diagram showing an outline of an embodiment of the present invention and a process of diagnosing an abnormality. In FIG. 1, 1 is a microphone, 2 is an amplifier, 3 is a high-frequency bandpass filter, 4
Is an envelope processing unit, 5 is a spectrum analysis unit, 6 is a spectrum level calculation unit, 7 is an abnormality determination reference value storage unit, 8 is an abnormality determination unit, 9 is a spectrum pattern calculation unit,
Is an abnormality type determination reference value storage unit, 11 is an abnormality type determination unit,
Reference numeral 12 denotes a determination result display unit, and reference numeral 13 denotes a determination result storage unit.
Note that the present invention is not limited to the example of FIG. 1, and each unit may be appropriately replaced with another known unit, and all the units are not essential and may be appropriately omitted in some cases. It is possible.

In the present embodiment, sound generated from a bearing to be diagnosed (not shown) is detected as sound pressure by the microphone 1, and signal level is amplified by the amplifier 2 in order to perform subsequent signal processing smoothly. Next, among the sound pressure signals,
The signal is passed through a high-frequency band-pass filter 3 that extracts only a signal in a high-frequency band including the ultrasonic band, and an envelope processing unit 4 generates an envelope signal of the sound pressure signal. The cutoff frequency of the high-frequency band-pass filter 3 is, for example, 10 kHz.
Set to. Further, the envelope processing is performed to convert the high-frequency amplitude modulation component into a low-frequency band and to facilitate the subsequent processing. Normally, the envelope processing is realized by passing the signal through an absolute value processing and then passing through a low-frequency band-pass filter.After that, the noise removal processing by the adaptive filter shown in Expressions (1) to (5) is added. By doing so, it is also possible to improve the SN ratio of the envelope signal.

[0017]

(Equation 1)

Here, y (t) is the output of the adaptive filter, ie, the envelope processing unit 4, x (t) is the input of the adaptive filter, ie, the output of the low-frequency band-pass filter, W (t) is the filter coefficient,
I is the tap length of the adaptive filter, μ is the step size parameter, γ is the leakage coefficient, Δ is a finite time delay, and e (t) is the error signal between the input x (t) and the output y (t) of the adaptive filter. is there.

The spectrum analysis unit 5 determines the spectrum distribution of the envelope signal, and the spectrum level calculation unit 6 determines the spectrum level, that is, the magnitude of the sound pressure, from the rms value of the spectrum waveform, for example. An abnormality presence / absence determination reference value storage unit 7 stores a determination reference value of a spectrum level. The reference value for determining the spectrum level may be either an absolute value of the sound pressure level or a relative value that is a ratio (abnormal value / normal value) of the abnormal and normal sound pressures. The abnormality presence / absence determination unit 8 compares the absolute value of the spectrum level calculated by the spectrum level calculation unit 6 or the relative value with the spectrum level in a normal state and the determination reference value stored in the abnormality presence / absence determination reference value storage unit 7. If the calculated spectral level is higher, the diagnosis target bearing is determined to be abnormal, and if it is lower, it is determined to be normal.

Next, if the abnormality presence / absence determination section 8 determines that there is an abnormality, the spectrum pattern calculation section 9 obtains an impact evaluation index when the spectrum distribution is regarded as a signal waveform. As the impact evaluation index, for example, the spectral crest factor SCF represented by the ratio of the maximum value SP to the effective value SR of the waveform shown in equation (6) and the absence of the statistical fourth moment shown in equation (7) There is a spectral kurtosis SK which is a dimension index.

[0021]

(Equation 2)

The abnormality type judgment reference value storage unit 10 stores judgment reference values relating to the impact characteristics of the spectrum distribution. The abnormality type determination unit 11 compares the impact evaluation index determined by the spectrum pattern calculation unit 9 with the determination reference value stored in the abnormality type determination reference value storage unit 10, and the determined impact evaluation index is larger. In this case, it is determined that a defect has occurred on the bearing rolling surface. Conversely, if the impact evaluation index obtained is smaller, it is determined that the bearing is poorly lubricated.

The judgment result regarding the presence / absence of abnormality and the type of abnormality is displayed on the judgment result display unit 12 together with the spectrum level and the value of the impact evaluation index. The judgment result storage unit 13 stores the contents displayed on the judgment result display unit 12 for each diagnosis together with the diagnosis date and time for the purpose of managing the tendency of the diagnosis data. The result stored in the determination result storage unit 13 can be arbitrarily displayed at a later date.

FIG. 4A shows a normal bearing (cylindrical roller bearing N).
315, 300 rpm) shows the spectral distribution of the sound pressure envelope signal obtained by analyzing the sound pressure at a position 4 m away from the apparatus by the method and apparatus of the present invention. As can be seen from the figure, the spectral level is low (spectral rms value: 0.04
8 mPa), and the spectral crest factor obtained for reference is 3.6, which is smaller than the abnormality type determination reference value (for example, 5.0).

FIG. 4 (b) shows the same bearing number (N315, 300 rpm) as in FIG. 4 (a) when the outer ring rolling surface has a scratch (abnormal initial stage) as in FIG. 4 (a). Shows the spectral distribution of Assuming that the spectrum level is high (spectral rms value: 0.317 mPa) and the presence or absence of abnormality is determined based on the relative value to the normal value (abnormal value / normal value), the relative value in this case is 6.7, and the presence or absence of abnormality is determined. Since the bearing is larger than the relative value determination reference value (for example, 4.0) stored in the reference value storage unit 7, this bearing can be determined to be abnormal. further,
Since the spectral crest factor is 7.0, which is larger than the above-described abnormality type determination reference value (5.0), it can be determined that the bearing has a defect.

FIG. 5 shows that the normal bearing is the same as that of FIG. 4A, the abnormal bearing is the same as that of FIG.
3 shows a plot of abnormal values / normal values obtained by the method and apparatus of the present invention when stepwise changing. FIG. 5 also shows plots of abnormal values / normal values according to the prior art (sound pressure level of 1 kHz to 5 kHz).

In the evaluation according to the prior art, the abnormal value / normal value is 1.3 to 2.0. Usually, it is necessary to set the abnormality determination reference value to at least about 2.0 to 4.0 in consideration of a change in measurement conditions, an error in the measurement value, and the like. . On the other hand, in the evaluation by the method and apparatus of the present invention, the abnormal value / normal value was 6.7 to
This is a sufficiently large value of 10.8, and the abnormality determination can be easily performed.

As can be seen from the above, by using the method for judging abnormality by the method and apparatus of the present invention,
It is understood that the abnormality of the bearing can be diagnosed at the initial stage.
In addition, diagnosis can be performed at a low rotational speed or a high rotational speed.

FIG. 6 plots the spectral crest factor values for the same abnormal bearing as FIGS. 4 (b) and 5 according to the method and apparatus of the present invention. At all rotation speeds, the spectral crest factors are all 7.0 or more, and it can be seen that the type of abnormality is not an abnormality such as poor lubrication but a defect such as a scratch. As can be seen from this figure, the type of abnormality can be specified by using the abnormality type determination method by the method and apparatus of the present invention without knowing detailed specifications of the bearing to be diagnosed.

Next, the envelope processing unit 4 performs absolute value processing of the sound pressure signal, passes through the low-frequency band-pass filter, and then passes the sound when passing through the adaptive filter using the least mean square algorithm as the adaptive algorithm. FIG. 7 (a) shows the spectrum distribution of the pressure envelope signal obtained for the normal and abnormal bearings similar to those in FIGS. 4 (a) and 4 (b).
(B). FIG. 8 plots outliers / normal values similar to those in FIG. 5 by the method and apparatus of the present invention, together with outliers / normal values according to the prior art. At any rotation speed, the abnormal value / normal value was as large as 8.7 to 16.2 in the evaluation by the method and apparatus of the present invention.
Abnormality determination can be made easier.

FIG. 9 also plots spectral kurtosis values for abnormal bearings with the method and apparatus of the present invention. At all rotation speeds, the spectral kurtosis is 19.
Since it is 3 or more, which is larger than the abnormality type determination reference value (for example, 10.0) related to this index, it can be seen that the type of abnormality is due to the occurrence of a defect such as a scratch.

As can be seen from the above, it is possible to easily diagnose an abnormality and an abnormality type of a bearing by using the abnormality determination method and the abnormality type determination method by the method and apparatus of the present invention.

[0033]

As is apparent from the above description, when the method and the apparatus of the present invention are used, the presence or absence and the type of the bearing are abnormal by performing the spectrum analysis of the envelope processing waveform of the high frequency component of the sound pressure. Can be diagnosed at an early stage. As a result, the diagnostic accuracy level of sound as a measurement parameter can be made closer to vibration, and the spread of equipment diagnosis can be greatly promoted.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an outline of an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of an octave analysis result of sound (sound pressure) generated from a bearing.

FIG. 3 is a diagram illustrating an example of a Fourier transform result of sound (sound pressure) generated from a bearing.

FIG. 4 is a diagram showing an example of a spectral distribution of a sound pressure envelope signal generated from normal and abnormal bearings by the method and apparatus of the present invention.

FIG. 5 is a diagram showing a comparison between a diagnostic parameter for performing abnormality determination using the method and apparatus of the present invention and a diagnostic parameter according to a conventional method.

FIG. 6 is a diagram showing diagnostic parameters for determining the type of abnormality by the method and apparatus of the present invention.

FIG. 7 is a diagram illustrating an example of a spectral distribution of a sound pressure envelope signal generated from normal and abnormal bearings by the method and apparatus of the present invention.

FIG. 8 is a diagram showing a comparison between a diagnostic parameter for performing abnormality determination using the method and apparatus of the present invention and a diagnostic parameter according to a conventional method.

FIG. 9 is a diagram showing diagnostic parameters for determining the type of abnormality by the method and apparatus of the present invention.

[Explanation of symbols]

 Reference Signs List 1 microphone 2 amplifier 3 high-frequency band-pass filter 4 envelope processing unit 5 spectrum analysis unit 6 spectrum level calculation unit 7 abnormality presence / absence determination reference value storage unit 8 abnormality presence / absence determination unit 9 spectrum pattern calculation unit 10 abnormality type determination reference value storage unit 11 Abnormality type judgment unit 12 Judgment result display unit 13 Judgment result storage unit

Claims (9)

[Claims] 1. A method for performing an abnormality diagnosis of a rolling bearing using sound, wherein sound generated from the rolling bearing is detected as a sound pressure, and only a sound pressure signal in a high frequency band including an ultrasonic band is extracted. The envelope signal is generated, the spectrum distribution of the envelope signal is obtained, the spectrum level is calculated, and the spectrum level is compared with a predetermined determination reference value. If the calculated spectrum level is large, the rolling bearing is determined to be abnormal. A method for diagnosing abnormalities of a rolling bearing. 2. For the rolling bearing determined to be abnormal by the abnormality diagnosis method according to claim 1, the spectrum pattern of the envelope signal is calculated, and when the spectrum pattern is smooth, it is determined that lubrication is defective. A method for diagnosing abnormalities of a rolling bearing, wherein it is determined that a defect has occurred when the spectrum pattern is steep. 3. The method according to claim 1, wherein a band of 10 kHz or more is adopted as the high frequency band. 4. A method of generating an envelope signal, comprising: performing an absolute value process on a sound pressure signal, performing a noise reduction process using an adaptive filter after passing through a low-frequency band-pass filter. Item 3. The abnormality diagnosis method for a rolling bearing according to item 1 or 2. 5. The method for diagnosing an abnormality of a rolling bearing according to claim 2, wherein an impact evaluation index of a spectrum waveform is adopted as the spectrum pattern. 6. An apparatus for diagnosing abnormalities of a rolling bearing using sound, comprising: a microphone for detecting sound pressure;
A high-frequency band-pass filter that extracts only the high-frequency band component of the detected sound pressure signal, an envelope processing unit that performs envelope processing on the sound pressure signal after passing through the filter, and a spectrum distribution of the obtained envelope signal. A spectrum analysis unit to be obtained, a spectrum level calculation unit to obtain the size of the spectrum distribution, an abnormality presence / absence determination unit that determines whether there is an abnormality by comparing the calculated spectrum level with a reference value, and a spectrum pattern for calculating a spectrum pattern An abnormality diagnosis device for a rolling bearing, comprising: a calculation unit; and an abnormality type determination unit that determines the type of abnormality. 7. An apparatus according to claim 6, wherein a filter having a cutoff frequency of 10 kHz is provided as the high-frequency band-pass filter. 8. The abnormality diagnosis device for a rolling bearing according to claim 6, wherein the envelope processing unit comprises an absolute value processing unit, a low-frequency band-pass filter, and an adaptive filter for removing noise. . 9. The abnormality diagnosis device for a rolling bearing according to claim 6, wherein a mechanism for calculating an impact evaluation index of a spectrum waveform is provided as the spectrum pattern calculation unit.