JPH0481655A - Method for specifying ae signal due to fatigue of bearing - Google Patents

Abstract

PURPOSE:To accurately and simply specify the frequency component of an AE signal due to the fatigue of a bearing by subtracting the frequency component detected on the basis of each of the AE signals of electric noise, mechanical noise, bearing slippage and bearing plastic deformation from the frequency component of the AE signal detected by the AE sensor arranged to a bearing. CONSTITUTION:The motor 21 and lubricant pump 23 of a lubricating drive apparatus are operated and the detection output through an AE sensor 28 and an A/D converter 31 is analyzed in frequency by a computer 27 to store the frequency component only of electric noise in the memory of the computer 27. The frequency component of the electric noise is subtracted from the frequency components obtained when all of the devices other than a hydraulic unit 24 of the lubricating drive apparatus are operated to calculate the frequency component only of mechanical noise to store the same in the memory. In the same way, the frequency components corresponding to bearing slippage and bearing plastic deformation are stored in the memory. Subsequently, the respective components stored in the memory are subtracted from the frequency components obtained when all of the devices of the lubricating drive apparatus are operated to accurately specify the AE signal due to the fatigue of a bearing.

Description

[Detailed description of the invention] [Industrial application field]

This invention deals with seven-course stays that occur in bearings due to bearing fatigue, which is a preliminary stage of bearing damage and destruction. The present invention relates to a method for identifying frequency components of an AE signal.

[Conventional technology]

Japanese Patent Application Laid-Open No. 62-282258 describes a method for predicting bearing life, that is, bearing damage and destruction, using AE in a frequency band that is thought to be related to changes in the internal structure of the bearing and the occurrence of internal cracks. A method using double signal is disclosed. There is no detailed explanation in the above publication about how to identify the AE multiplied signal due to the occurrence of internal cracks in the bearing, that is, bearing fatigue. We detect all AE that occurs during the life of the bearing, conduct frequency analysis and amplitude measurements at various stages, cut the bearing and check the internal AE, and identify the AE multiplied signal frequency component generated by the bearing itself. AE due to bearing fatigue can be avoided by associating with bearing legal experts.
This specifies the double signal frequency.

[Invention or problem to be solved]

However, if an attempt is made to identify the AE multiplied signal due to bearing fatigue using the above-mentioned conventional method, it is necessary to cut the bearing and check the internal condition of the bearing, which is time consuming and costly. Furthermore, the generated AE multiplied signal includes various AE multiplied signals other than the AE signal due to bearing fatigue.
Frequency analysis becomes complicated, and depending on how the data is collected, it may become uncertain whether the frequency components can be identified.
There is. SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a method for specifying an AE multiplied signal due to bearing fatigue, which can accurately and easily specify the frequency component of an AE multiplied signal due to bearing fatigue.

[Means to solve the problem]

In order to achieve the above object, a method for specifying an AE multiplied signal due to bearing fatigue according to claim 1 of the present invention provides an AE multiplied signal identification method due to fatigue of a flIl bearing that supports a shaft rotated by a rotary drive device.
A method for identifying a signal, the A
From the AE multiplied signal frequency component detected by the E sensor, electrical noise\frequency component detected by the E signal detection means, mechanical noise frequency component detected by the AE signal detection means, bearing slip, AE signal detection means and the frequency component detected by the bearing plastic deformation AE signal detection means are respectively subtracted by the subtraction means to obtain the AE multiplied signal frequency component due to bearing fatigue. Due to bearing fatigue in Section 2.3. To identify the E signal, use a testing machine as shown in Figure 1. This testing machine includes a bearing rotation device, a bearing lubrication drive device, and a control device that controls the lubrication drive device. Further, the bearing lubrication drive device includes a motor, an electromagnetic coupling that connects the motor and the bearing rotation device, a lubricating oil pump, and a valve provided between the lubricating oil pump and the bearing rotating device, It has a hydraulic unit that applies a load to the bearing. Furthermore, the test machine includes a sensor for detecting the AE multiplied signal, an amplifier for amplifying the AE multiplied signal detected by the sensor, an A/D converter, and an AE multiplied signal A/D converted by the A/D converter. It is equipped with a computer that analyzes frequencies. In the above testing machine, the AE multiplied signal generated under various conditions is detected by the sensor, amplified by an amplifier, tintalized by an A/D converter, and frequency analyzed by a computer. The method for identifying an AE multiplied signal due to bearing fatigue according to claim 2 of the present invention uses the test machine having the above configuration, and first operates only the motor and the lubricating oil pump of the bearing lubrication drive device to detect the AE signal caused by electrical noise. After generating a double signal, this A
E-multiply signal frequency analysis is performed to find the frequency component of electrical noise. Next, the valve of the bearing lubrication drive device is opened and the motor, lubricating oil pump, and electromagnetic coupling are operated to detect AE consisting of electrical noise and mechanical noise. After generating the double signal, find the frequency component of this AE double signal, and then
The frequency component of the electrical noise is subtracted from the frequency component of the E-fold signal to obtain the frequency component of only the mechanical noise. Next, the valve of the bearing lubrication drive device is idled, and the motor, lubricating oil pump, and electromagnetic coupling are Electrical noise and machine noise caused by sliding of bearings during operation
After generating the A E signal consisting of the E times signal, this A
Find the frequency component of the E signal, then subtract the frequency component of the electrical noise and the frequency component of the mechanical noise from this AE multiplied signal frequency component to find the frequency component of only the AE signal due to bearing slippage, Next, among the bearing lubrication drive devices mentioned above, there is an oil tapping unit. The bearing is operated to plastically deform the bearing, and the AE multiplied signal frequency component due to the plastic deformation of the bearing that occurs at this time is determined. Next, the entire bearing lubrication drive system is operated so that an appropriate load is applied to the bearing. , the frequency component of the AE multiplied signal generated at this time is determined, and from the frequency component determined when the entire bearing lubrication drive device is operated, the AEC signal due to the electrical noise, mechanical noise, bearing slippage, and plastic deformation of the bearing is determined. The purpose of this method is to subtract each frequency component of the XC signal to obtain the AEC signal frequency component due to bearing fatigue. The AEC signal identification method due to bearing fatigue according to claim 3 of the present invention uses the testing machine configured as described above to first close the valve of the bearing lubrication drive device and operate the motor, lubricating oil pump, and electromagnetic coupling. AEC consists of electrical noise, mechanical noise, and AC signals generated from bearing slippage.
After the signal is generated, this AC signal is frequency-analyzed to find the frequency component.Next, only the hydraulic unit of the bearing lubrication drive device is operated to plastically deform the bearing, and the bearing deformation that occurs at this time is Determine the AEC signal frequency component due to plastic deformation, then operate the entire lubrication drive device so that an appropriate load is applied to the bearing, determine the frequency component of AE (No. 3) that occurs at this time, and operate the entire lubrication drive device. From the frequency component obtained in the case of
It is characterized by determining the frequency component of AE due to bearing fatigue.

[Examples] The present invention will be described in detail below with reference to illustrated examples. Figure 1 shows how different types of AC signals are generated and the AE
This is a test machine that analyzes C signal frequency, and in this figure,
21 is a motor, 22 is an electromagnetic coupling, 23 is a lubricating oil pump, 24 is an oil stamp unit, 32 is a valve,
These devices constitute the lubricated drive for the bearing rotation device 25 during the test number. Each device of this lubricated drive device is
It is surrounded by a dashed line in FIG. 1 and is controlled by a control unit 26 which is controlled by a conobulator 27. 28 is an AE processor that detects the AEC signal. The E signal detected by the sensor 28 is first amplified by a preamplifier 29, then by a main amplifier 30, and then amplified by an analog/digital (A/D) converter 3.
1 to convert it into a tit/tal signal. A/D converted A
The frequency of the EC signal is analyzed by the computer 27. The frequency component of the AEC signal that is correlated with structural changes inside the bearing and the occurrence of internal cracks, that is, the frequency component of the AEC signal generated from the bearing when fatigue occurs due to rolling of the bearing, is specified by the following method. Note that in each of the following steps, AEC signal detection, amplification, A/I) conversion, and frequency analysis are performed in the same way. As shown in FIG. 4, first, in step S1, under the control of the control unit 26, the motor 21 and the lubricating oil pump 23 of each device of the lubricating drive device are operated, and the frequency components of the electrical noise are determined. Under this condition, only electrical noise is generated, so the frequency components obtained through AE frequency analysis by the computer 27 are frequency components of only electrical noise. The power spectrum of electrical noise is shown in FIG. 2A. The power spectrum of the electrical noise is stored as a tee/tal motor E in the memory (not shown) of the computer 27. Next, in step S2, the frequency component of mechanical noise is determined as follows. This station. Of all the devices in the lubrication drive system, this is the hydraulic unit. Activate all devices outside the tiger. In this case, not only mechanical noise but also electrical noise will be generated. FIG. 2B shows a power spectrum containing a mixture of electrical and mechanical noise. Therefore, by subtracting the frequency component of the electrical noise stored in the memory as the motor E from the frequency component obtained by analyzing the remaining number of laps by the cono viewer 27, the frequency component of only the mechanical noise is obtained. It is stored in memory as the motor M of the tee/tal. Next, in step S3, frequency components of AE caused by bearing slippage are determined. The motor 21, electromagnetic force, pull 22 and lubricating oil pot 23 are activated to cause the bearing to slide. At this time, the valve 32 is kept closed. As a result, lubricating oil is not supplied to the bearings, causing slippage in the bearings. It includes frequency components of the AE multiplied signal, electrical noise, mechanical noise, and AE caused by bearing slippage that occur at this time. Therefore, the frequency components of electrical noise and mechanical noise stored in the memory as data E and M are subtracted to obtain only the frequency component of AE due to slip, and this is stored in the memory as the static motor S. Next, in step S4, only the hydraulic unit 24 of the j lubrication drive device is operated. When the hydraulic unit 24 is operated to gradually increase the oil pressure, plastic deformation occurs in the bearing. Therefore, the AEs that occur under these conditions
The double signal is due only to the plastic deformation of the bearing (2nd D
(see figure p). Therefore, the frequency components obtained here are stored as data P corresponding to plastic deformation. Second
As shown in Figure D, the AE due to plastic deformation of the bearing is 10
It has characteristics in the frequency band of 0 KHz = 150 KHz and the frequency band of 300 KHz to 450 KHz. Next, in step S5, all devices of the lubricating drive device are activated. At this time, it is desirable to apply a high load, that is, a high force, to the bearing by the hydraulic unit 24. It grows under these conditions. A E (@ indicates A caused by falling fatigue.
E signal is not high enough, AE multiplication signal caused by electrical noise, mechanical noise, shear, and A caused by plastic deformation.
It also includes the E signal (see Figure 3A). Therefore, the AE multiplied signal caused by electrical noise, mechanical noise, and slip, and the AE multiplied signal frequency components E, M, S, and P caused by plastic deformation are subtracted from the frequency components obtained here. As a result, as shown in FIG. 3B, an AE multiplied signal frequency component related to rolling fatigue is obtained. From the power spectrum shown in FIG. 3B, it can be seen that AE caused by rolling fatigue is characterized by frequency components in the band of 150 KHz to 300 KHz. Here, each of the above steps Sl, S2. S3.
The lubrication drive devices operated in S4 and S5 are summarized in Table 1 below. Due to slipping, on-on-on closed AE generated off. Due to plastic deformation, off-off off closed. Due to on-fatigue, on-on-on opened. Valve 24/Hydraulic unit on topsoil According to the above method, the AE multiplied signal frequency components caused by electrical noise, mechanical noise, and slip are individually determined and individually stored in memory. However, for example, as shown in the flowchart in Figure 5,
Three types of frequency components E, λ1.AE multiplied signals due to electrical noise, mechanical noise, and slip. S may be calculated simultaneously. In this case, step Sll in FIG.
SI2 and S13 respectively correspond to step S3 in FIG.
, S4, and S5. By the way, destruction or damage to a bearing is determined by measuring the AE multiplication signal due to bearing fatigue using a double method.
It can be predicted by detecting the AE multiplied signal in the 00KHz frequency band. FIG. 1 shows a device for predicting the life of such a bearing. In Fig. 6, 1 is a sensor attached to the bearing and detects the acoustic and quemino/yon signals from the bearing;
2 is a preamplifier, 3 is a bandpass filter that passes an AE multiplied signal in the band of 150 KHz to 300 KHz, which is specified by the above method and represents bearing fatigue, 4 is a main amplifier, and 5
6 is an envelope detection circuit, 6 is an A/D converter, 7 is an AE multiplied signal inputted from the A/D converter 6, and is compared with the reference value. If the above signal exceeds the reference value, it is detected as a sign of destruction. It is a computer as a calculation device that detects the situation and issues an alarm. With the above configuration, the AE multiplied signal detected by the sensor 1 is input to the band pass filter 3 via the preamplifier 2, and only frequency components in the range of 1.150 to 300 KHz are extracted. This output is further amplified by the main amplifier 4, envelope detected by the envelope detection circuit 5, A/D converted by the A/D converter 6, and input to the computer 7. The computer issues an alarm when this output exceeds the standard value. In this way, by distinguishing among the AE multiplied signals generated by the bearing, the AE multiplied signals that have frequency components (150 to 300 KHz) that are correlated with internal structural changes and the occurrence of internal cracks, the bearing can be easily and accurately detected. It is possible to predict the occurrence of cracks. In the above embodiment, the computer 7 compares the signal from the A/Di converter 6 with the reference value, and determines whether a sign of destruction has been detected, the number of AE multiplied signals, the waveform shape, the amplitude, the state of occurrence, etc. Signs of destruction may be detected based on comprehensive judgment. FIG. 9 shows another device for predicting bearing life or failure. In this Figure 9, 11.11'
, 11°°- is the A/E sensor, 12. ]2”, 12
"・ is preamplifier, 13 is preamplifier 12.12",
12"..., a multiplexer for switching 4, a band pass filter that removes noise from the AE multiplied signal, 15 a main amplifier, 16 an A/D converter, 17 a computer as an arithmetic device including discrimination means, and 18 an AE multiplexer. A memory 19 is a display device for displaying the AE multiplied signal waveform, and the computer 7 outputs a signal to control the multiplexer 13 and bandpass filter 14. , the operation of the AE destruction prediction device is described in Section 1011.12.
The flowchart shown in the figure will be explained. When the computer 17 is operated, a measurement routine (S,) shown in FIG. 10 is performed. As shown in step S4+ in FIG. 11, this measurement routine first switches the multiplexer 13 to the AE sensor 11.
Select l I'. Next, proceeding to step S4+, the band of the bandpass filter 14 is selected depending on the type of the bearing to be measured and the mounting location. Next, the process proceeds to step S43, where the AE multiplied signal waveform is stored in the memory 18. In step S44, it is determined whether steps S41+S42+S43 have been performed for all AE sensors. If not, the process returns to S41; if the process has been performed, the process advances to step S45. Step S
45, based on the previously memorized waveform take,
Perform parameter calculations as shown in Figure 2. This parameter calculation is performed as shown in FIG.
Calculation of the occurrence rate (total number of occurrences) that counts the number of times the E times signal level exceeds a certain value (S52), calculation of the AE times signal amplitude in a specific band (S53), AE times signal step, as shown in step Sol. After detecting the envelope, calculation of the duration at which the level exceeds a predetermined level value (S 5), and calculation of frequency components by fast Fourier transform of the AE multiplied signal (S s5).
) etc. When the measurement routine (Sl) is completed, the process proceeds to step S in FIG. 10, where it is determined whether the parameters obtained from the parameter calculation are stable. If it is not stable, the process returns to step S1, and if it is stable, the process proceeds to step S3, where the above parameters are multiplied by a constant to set a reference value. By doing this, the reference value of the E signal for each bearing being tested is appropriately set. Next, in order to actually predict destruction, step S4
Then, the measurement routine shown in FIG. 11 is performed again to calculate the parameters, that is, the incidence, amplitude, and duration of one frequency component. Next, proceed to step S5, and
The reference value obtained in step 3 is compared with the parameter obtained in step S4, and if all the parameters mentioned above are smaller than the reference value, the process returns to step S4, and if all the parameters are larger than the reference value, the process is destroyed. It is determined that this is a sign of a problem occurring, and the process proceeds to step S6, where an alarm is issued. In step S7, a waveform is displayed on the surface water device 19 based on the waveform data stored in the memory 18 according to the operator's instructions. The operator compares the waveform with the waveform at the time of appropriate prediction in the past, and if it is similar to that, determines that it is a sign of destruction, and if not, causes the computer to return to step S4,
Continue the determination process. In this way, instead of issuing an alarm by comparing parameters with reference values, the AE multiplied signal waveform is displayed on the display device 19 and the suitability of the prediction is confirmed based on the entire waveform, so it is possible to predict destruction. can be done reliably. In the AE destruction prediction device having the above configuration, in step S5 constituting the discriminating means, the reference values of the occurrence rate, amplitude, and duration frequency components and the newly determined parameters of the occurrence rate, amplitude,
All duration 1 frequency components may be compared to identify a precursor to rupture, or a combination of one or more of these may be compared to identify a precursor to rupture. In addition, this AE destruction prediction device is equipped with a titanized AE
Since the processing is carried out by a discriminating means composed of multiplication signal software, discrimination for various parameters can be easily and inexpensively performed.

【Effect of the invention】

As is clear from the above, the method for identifying the AE signal due to bearing fatigue according to claim 1 of the present invention is not suitable for identifying the AE multiplied signal due to fatigue of the bearing that supports the shaft rotated by the rotary drive ball device. and was placed on the bearing. From the frequency component of the AE signal detected by the AE sensor, the frequency component detected by the electrical noise AE signal detection means, the frequency component detected by the mechanical noise AE signal detection means, and the frequency component detected by the bearing slip AE signal detection means. The detected frequency component and the frequency component detected by the bearing plastic deformation AE signal detection means are each subtracted by the subtraction means to obtain the frequency component of the AE multiplied signal due to bearing fatigue, so the AE multiplied signal due to bearing fatigue is accurately calculated. Can be specified. Further, according to the method for specifying an AE signal due to bearing fatigue according to claim 2 of the present invention, the frequency components of the AE multiplied signal caused by electrical noise, mechanical noise, and bearing slippage are obtained, respectively. Furthermore, by generating an AE signal consisting of AE multiplied signals caused by these four types of AE multiplied signals and bearing fatigue, and subtracting the above four types of AE multiplied signal frequency components from the total frequency components of this AE multiplied signal, Since only the frequency component of the AE multiplied signal is determined, the AE multiplied signal due to bearing fatigue can be accurately specified. Moreover, if the testing machine according to the present invention is used,
You can easily create conditions for generating each AE multiple signal,
Each AE multiplied signal frequency component such as electrical noise can be easily obtained. Therefore, the AE multiplied signal frequency component due to bearing fatigue can be easily identified. Furthermore, this method does not require cutting the bearing, so it can be performed at low cost. Further, according to the method of claim 3 of the present invention, electrical noise,
Since the AE multiplied signal frequency components generated by mechanical noise and bearing slip are collectively determined, the number of steps can be reduced compared to the method of claim 1.

[Brief explanation of drawings]

FIG. 1 is a flowchart of a testing machine used to carry out the method of identifying AE multiplied signals due to bearing fatigue according to the present invention. This is a diagram. FIG. 2A is a diagram showing a power spectrum of an AE multiplied signal consisting of electrical noise. FIG. 2B is a diagram showing an AE multiplied signal power spectrum consisting of electrical noise and mechanical noise. FIG. 2C is a diagram showing the power spectrum of the AE multiplied signal, which is composed of the AE multiplied signal due to electrical noise, mechanical noise, and bearing slippage. FIG. 2D is a diagram showing the power spectrum of the AE signal due to plastic deformation of the bearing. Figure 3A shows AE caused by electrical noise, mechanical noise, and slippage.
Double signal AE due to plastic deformation Double signal and A due to bearing fatigue
FIG. 3 is a diagram showing the frequency analysis result of an AE multiplied signal consisting of an E multiplied signal. FIG. 3B is a diagram showing the AE multiplied signal power spectrum due to bearing fatigue. FIGS. 4 and 5 are flowcharts each showing a method for specifying the AE multiplied signal frequency component due to bearing fatigue. FIG. 6 is a block diagram of a bearing life prediction device that utilizes the frequency band of AE) No. 1 due to bearing fatigue identified by the above-mentioned identification method. FIG. 7 and FIG. 8 are diagrams showing the crystal structure of the bearing member. FIG. 9 is a block diagram of an AE fracture prediction device that utilizes the AE multiplied signal frequency band due to bearing fatigue specified by the above specification method. FIGS. 10 and 11 are flowcharts showing computer Arfnosm. FIG. 12: A flowchart showing parameter calculation. 2I Motor, 22 Electromagnetic coupling, 23 Lubricating oil pump, 24 Hydraulic unit, 25 Bearing rotating device, 26 Control unit, 32 Valve. Patent applicant: Koyo Seiko Co., Ltd. Agent
Patent Attorney Blue and White Haka1 Figure 2A Figure 2B (KH2) Figure 2C Figure 2D Number of Peripherals (KHz) Figure 4 Figure 3A Figure 3B Number of Peripherals (KHz) Figure S Figure 7 Figure 8 Figure 0 Figure

Claims (3)

[Claims] (1) A method for identifying an AE signal due to fatigue of a bearing that supports a shaft rotated by a rotary drive device, the A signal being detected by an AE sensor placed on the bearing.
From the frequency components of the E signal, the frequency components detected by the electrical noise AE signal detection means, the frequency components detected by the mechanical noise AE signal detection means, and the bearing slip AE
The frequency component detected by the signal detection means and the frequency component detected by the bearing plastic deformation AE signal detection means are each subtracted by the subtraction means to determine the A due to bearing fatigue.
A method for identifying an AE signal due to bearing fatigue, the method comprising determining the frequency component of the E signal. (2) a bearing rotating device, a motor, an electromagnetic coupling connecting the motor and the bearing rotating device, a lubricating oil pump, a valve provided between the lubricating oil pump and the bearing rotating device, and a bearing. a bearing lubrication drive device having a hydraulic unit that applies a load to the bearing lubrication drive device; a control device that controls each device of the bearing lubrication drive device; a sensor that detects an AE signal; and an amplifier that amplifies the AE signal detected by the sensor. , Using a testing machine equipped with an A/D converter and a computer that analyzes the frequency of the AE signal A/D converted by the A/D converter, first, the motor and lubricant of the bearing lubricating drive device were tested. After operating only the pump to generate an AE signal consisting of electrical noise, this AE signal is frequency-analyzed to determine the frequency component of the electrical noise. Next, the valve of the bearing lubrication drive device is opened, and the motor, After operating the lubricating oil pump and electromagnetic coupling to generate an AE signal consisting of electrical noise and mechanical noise, the frequency component of this AE signal is determined, and then the frequency component of the electrical noise is determined from the frequency component of this AE signal. Next, the valve of the bearing lubrication drive device mentioned above is closed, and the motor, lubricating oil pump, and electromagnetic coupling are operated to eliminate electrical noise, mechanical noise, and bearing slippage. A caused by
After generating an AE signal consisting of a AE due to slippage
Find the frequency component of only the signal, then operate only the hydraulic unit of the bearing lubrication drive device to plastically deform the bearing, find the frequency component of the AE signal due to the plastic deformation of the bearing that occurs at this time, and then , operate the entire bearing lubrication drive system so that an appropriate load is applied to the bearing, find the frequency component of the AE signal generated at this time, and from the frequency component found when the entire bearing lubrication drive system is operated, calculate the above. AE caused by bearing fatigue is calculated by subtracting the frequency components of electrical noise, mechanical noise, AE signal caused by bearing slippage, and AE signal caused by plastic deformation of bearing.
A method for identifying an AE signal due to bearing fatigue, the method comprising determining the frequency component of the signal. (3) a bearing rotating device, a motor, an electromagnetic coupling connecting the motor and the bearing rotating device, a lubricating oil pump, a valve provided between the lubricating oil pump and the bearing rotating device, and a bearing. a bearing lubrication drive device having a hydraulic unit that applies a load to the bearing lubrication drive device; a control device that controls each device of the bearing lubrication drive device; a sensor that detects an AE signal; and an amplifier that amplifies the AE signal detected by the sensor. , Using a testing machine equipped with an A/D converter and a computer that analyzes the frequency of the AE signal A/D converted by the A/D converter, first, close the valve of the bearing lubrication drive device, AE caused by electrical noise, mechanical noise, and bearing slippage when operating motors, lubricating oil pumps, and electromagnetic couplings.
After generating an AE signal consisting of a signal, frequency analysis is performed on this AE signal to find the frequency component.Next, only the hydraulic unit of the bearing lubrication drive device is operated to plastically deform the bearing, and the AE signal generated at this time is Find the frequency component of the AE signal due to plastic deformation of the bearing, then operate the entire lubrication drive device so that an appropriate load is applied to the bearing, find the frequency component of the AE signal generated at this time, and calculate the frequency component of the AE signal generated at this time. The frequency component of the AE signal consisting of the electrical noise, mechanical noise, and AE signal due to bearing slippage, and the frequency component of the AE signal due to plastic deformation of the bearing are subtracted from the frequency component obtained when the bearing is operated. A method for identifying an AE signal due to bearing fatigue, the method comprising determining the frequency component of an AE signal due to bearing fatigue.