JPS6078343A - Detection for rubbing of rotary machine - Google Patents

Abstract

PURPOSE:To detect rubbing with high sensitivity by providing an acoustic sensor in a bearing part of a rotary machine to detect high-frequency abnormal sounds which are generated when rubbing occurs. CONSTITUTION:Acoustic sensors 5a and 5b are provided on sliding bearings 2a and 2b of a rotor 1. High-frequency abnormal sounds due to rubbing which occurs in the position marked with a black star are propagated in the rotor 1 and are detected by acoustic sensors 5a and 5b. Their outputs pass amplifiers 6a and 6b, filters 7a and 7b, and detecting circuits 8a and 8b, and signals f(t) and g(t) are inputted to a cross-correlation function analyzer 9, and a cross-correlation function is calculated and is sent to a monitor 10. Thus, presence/absence of rubbing is detected with a high sensitivity.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention detects high-frequency abnormal gold generated when a rotating part of a rotating machine such as a steam turbine or a turbine generator causes a rubbing phenomenon during rotation. Accordingly, the present invention relates to a rubbing detection method for a rotating machine suitable for detecting the occurrence of rubbing.

[Background of the invention]

When rubbing occurs in rotating machinery, it not only causes abnormal vibrations of the machinery, but also causes problems in operation.
It is known to be dangerous due to the force generated in the event of a flying machine rotor.

Conventionally, a method of detecting rubbing based on the amount of vibration change in the bearing has been used, but it has been difficult to detect because of poor sensitivity. In addition, a method is used in which an acoustic sensor (mainly an acoustic emission sensor) is installed in a stationary part of a rotating machine, and the occurrence of silabitz is detected by the amplitude change of the high-frequency abnormal sound signal that occurs when rubbing occurs. Although it is possible to detect rubbing at low speed rotation, it has the disadvantage that it becomes impossible to detect when the rotation speed increases and background noise increases or when the rubbing is slight.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide a rubbing detection method that is not only capable of identifying the presence or absence of rubbing even when background noise is large, but is also capable of identifying the presence or absence of rubbing even when the rubbing is slight.

[Summary of the invention]

A feature of the present invention is that an acoustic sensor is installed in the bearing of a rotating machine, the acoustic sensor detects the high-frequency abnormal sound that occurs when rubbing occurs, and this signal is processed so that error rubbing can be detected with high sensitivity. At the point.

[Embodiments of the invention]

A rubbing detection method according to an embodiment of the present invention will be explained below with reference to FIG. In a rotating machine, a rotor 1 is held by beveled bearings 2a and 2b, and a casing 3
covered by. Hypothetically, if rubbing occurs at the rubbing occurrence location 4 shown in the same figure, a high-frequency abnormal sound signal (hereinafter referred to as an acoustic signal) due to the rubbing will propagate within the rotor 1 and cause damage to the sliding bearings 2a, 2b and the rotor 1. The lubricating oil film formed between the sliding bearings 2a, 2. b. In order to detect the acoustic signal, as shown in the same figure, multiple bearing parts (bearings, or housing parts are also possible) 2a, 2b are used.
Acoustic sensors 5a and 5bf are installed at the respective locations. Next, the outputs of the acoustic sensors 5a and 5b are amplified by amplifiers 6a and 6b, respectively, and then passed through filters 7a and 7b to remove signals with unnecessary frequency components. Further, the signals processed by the filters are detected by detection circuits f3a and 8b, and then each signal is input to a cross-correlation function analyzer 9. The signal '! output from the detection circuit 8a!
i-f (t), the signal output from the detection circuit 8b is g
(t), the cross-correlation function analyzer 9 calculates the following equation.

C, (τ) = J(t)・g (10τ) ・・・・・・(
1) C, (t): Cross-correlation function τ: Time lag - in equation (1) means time average.

That is, in the cross-correlation function analyzer 9, the data value f (t) at a certain time t and t
The temporal relationship with the data value g (10τ) output from the detection circuit 8b at a time delayed by τ is examined.

Next, the calculation results of the cross-correlation function analyzer 9 are input to the monitor 10, and the results are displayed.

The above-mentioned processing method will be specifically explained using FIG. 2.

This figure shows an example of the output of each circuit shown in FIG. 1 in a steam turbine when there is no rubbing, when the rubbing is slight, when the rubbing intensity is medium, and when the rubbing is large.

The output waveforms of the amplifiers 6a and 6b are almost indistinguishable from those with no rubbing to those with large rubbing.

This is because the noise caused by the rotation of the steam turbine and the steam noise are large, so that the acoustic signal caused by rubbing is buried. By passing the aforementioned signals through filters 7a and 7b and then detecting them, an output waveform as shown in the figure is obtained.

If the rubbing is large, a slightly periodic sine wave can be seen at the top of the detected waveform. This is because when the rubbing occurs, the acoustic signal due to the rubbing is modulated into background noise, and this signal appears in a small amount when detected. However, when the rubbing is moderate or less, the intensity of the acoustic signal due to the rubbing is small, that is, the modulation rate is small, so as shown in the figure, no change is observed in the detected waveform. Therefore, after the above-mentioned detected waveform is processed by the cross-correlation function analyzer 9, it is displayed on the monitor 10 as shown in the figure.

The rubbing phenomenon is characterized by occurring periodically once per rotation of the rotor of a rotating machine, and when there is a strong periodic correlation (for example, when the 2 pings in the same figure are large) (for example, when the sine waves It is similar to the calculation result of the cross-correlation function (correlation). Finally, a periodic (temporally related) sine wave signal included in the detected waveform mentioned above has been detected.
The presence or absence of rubbing can be determined from the display example on the monitor 10. Further, even when there is slight or moderate rubbing in which no change is observed in the detected waveform, correlation processing is similarly performed to obtain the display results shown in the figure, so it is possible to determine the presence or absence of rubbing. This means that even if the rubbing is light or moderate, the acoustic signal caused by the rubbing is slightly modulated by background noise, so by performing correlation processing, periodic acoustic signals can be detected. It is.

On the other hand, if there is no rubbing, it is just random noise, and as shown in the figure, no processing results specific to the occurrence of rubbing can be obtained. In addition, the period T (period of the sine wave signal) shown in the monitor display example in Fig. 2 also corresponds to the rotation period of the rotor, so by measuring T, it is possible to detect other periodic disturbance noises, such as bearing damage. It can also be distinguished from acoustic signals generated by

As explained above, by using this method, it is not only possible to detect rubbing even when the background noise is large, but also it is possible to detect when the rubbing is slight, and it is also indistinguishable from periodic noise caused by external disturbances. As a result, accidents in rotating machinery can be prevented, which is an extremely significant industrial effect.

Next, the third embodiment of the rubbing detection method according to the present invention will be described.
Let me explain to you. An acoustic sensor 22 is installed on a sliding bearing (or a bearing or housing) 21 that supports a rotor 20 of a rotating machine. Next, the output total amplifier 2 of the acoustic sensor 22
After being amplified in step 3, the signal is passed through a filter 24 to remove signals with unneeded frequency components. Furthermore, the signal processed by the filter 24 is input to the detection circuit 25, and the detection circuit 25
The detected signal output from the autocorrelation function analyzer 26 is inputted to the autocorrelation function analyzer 26.

When the detection signal output from the detection circuit 25 is f(t), the autocorrelation function analyzer 26 calculates the following equation.

C, (s=f(t)・fct+τ) ・・・・・・(2
)C, (τ): autocorrelation function τ: time lag - in equation (2) means time average.

In other words, in the autocorrelation function analyzer 26, the data value f (t) at the time t, which is output multiple times from the detection circuit 25, is
The temporal relationship between the data value f (10τ) at a time delayed by 97 hours from t is examined. Next, the calculation result of the autocorrelation function analyzer 26 is input to the monitor 27, and the result is displayed.

The above-mentioned processing method will be specifically explained using FIG. 4.

This figure shows an example of the output of each circuit shown in FIG. 3 when there is no rubbing in the steam turbine, when the rubbing is slight, when the rubbing intensity is medium, and when the rubbing is large. A detailed explanation of the output waveforms of the amplifier 23 and the detection circuit 25 is the same as that explained with reference to FIG. 2, so a detailed explanation will be omitted here.

The detected waveform shown in FIG.
After processing, the image is displayed on the monitor 27 as shown in the same figure. As mentioned above, the rubbing phenomenon is characterized by occurring periodically once per rotor rotation, and as shown in the detected waveform in the case of large rubbing in the same figure, periodic signal components corresponding to the rotation speed are generated. Since it is modulated and included in the background noise, the result is similar to the calculation result of, for example, a sine wave autocorrelation function. Finally, a signal component with periodicity (temporally related) included in the above-mentioned detected waveform is detected, and the presence or absence of rubbing can be determined from the display example on the monitor 27. In addition, even in the case of slight or moderate rubbing where no change was observed in the detected waveform, the acoustic signal due to rubbing is modulated by background noise.
Since the acoustic signal has periodicity, by similarly performing autocorrelation processing, the processing results shown in the figure can be obtained, so that it is possible to determine the presence or absence of rubbing.

- If there is no rubbing, it is just random noise, so as shown in the figure, a processing result specific to the occurrence of rubbing cannot be obtained. ti, the period T of the sine wave signal in the monitor display example shown in Fig. 4 also corresponds to the rotation period of the rotor, so by measuring T, it is possible to detect other periodic disturbance noise caused by damage to bearings, etc. It can also be distinguished from acoustic signals.

〔Effect of the invention〕

As explained above, by using this method, it is possible not only to detect rubbing even when the ground noise is large, but also when the rubbing is slight, and it can also be distinguished from periodic noise caused by disturbances. This has an extremely significant industrial effect in preventing accidents in rotating machinery.

[Brief explanation of the drawing]

Fig. 1 is a configuration diagram showing an embodiment of the rubbing detection method for a rotating machine according to the present invention, Fig. 2 is a waveform diagram for explaining the embodiment of Fig. 1, and Fig. 3 is a diagram showing another embodiment of the present invention. A configuration diagram showing an example, and FIG. 4 is a waveform diagram for explaining the embodiment of FIG. 3. 1...Rotor, 2ai 2b...Slide bearing, 3.
...Kezong, 4...Rubbing occurrence point, ja, 5
b...Acoustic sensor, 6a, 6b...Amplifier, 7a,
7b... Filter, 8a, 8b... Detection circuit, 9.
... Cross-correlation function analyzer, 10... Monitor, 26...
・Autocorrelation function analyzer. ¥10

Claims (1)

[Claims] 1. In a method of detecting sliding between a rotating part and a stationary body in a rotating machine, that is, a rubbing phenomenon using an acoustic sensor, a first acoustic sensor and a first acoustic sensor are provided in a bearing part of the rotating machine. a first means for installing a second acoustic sensor in a bearing portion different from the first acoustic sensor, and amplifying and detecting the signal received by the first acoustic sensor to obtain a detected waveform;
a second means for amplifying and detecting the signal received by the acoustic sensor to obtain a detected waveform; and a second means for obtaining a detected waveform by amplifying and detecting the signal received by the acoustic sensor; A rubbing detection method for a rotating machine characterized by detecting rubbing from a calculation result of a cross-correlation function.