JP5740208B2 - Bearing diagnosis method and system - Google Patents

Description

  The present invention relates to a method and system for diagnosing the life of a bearing of a rotary machine using acoustic emission.

  The vibration method and the acoustic emission method (AE method) are often used for bearing diagnosis. For example, Japanese Patent Application Laid-Open No. 2004-233284 [Patent Document 1] discloses an invention relating to a “rolling bearing unit diagnostic device and method” that uses a vibration sensor, an acoustic emission sensor (referred to as an AE sensor in the present specification) and the like as detection means. Is disclosed. Japanese Patent Application Laid-Open No. 2002-188411 [Patent Document 2] and Japanese Patent Application Laid-Open No. 8-210347 [Patent Document 3] also disclose an invention relating to an “abnormality diagnosis apparatus” for diagnosing a bearing abnormality using an AE sensor, An invention relating to “a damage detection method for an ultra-low speed rotating bearing” is disclosed.

JP 2004-233284 A JP 2002-188411 A JP-A-8-210347

  In many of the conventional techniques, diagnosis or evaluation is performed based on whether or not a characteristic frequency depending on the bearing appears by analyzing the magnitude of the measurement value and the period. Since both the vibration method and the AE method have different operating conditions depending on the target device, a management standard is set uniformly for each product, and the relative evaluation with the normal time is performed. For this reason, there is a possibility that the signal may change depending on the installation state of the sensor and the operation state of the target device, which causes a problem in measurement accuracy. Moreover, since it was relative evaluation, there was a problem that the normal state could not be grasped with the first target device.

  The vibration method is general, but is difficult to evaluate unless the damage is macroscopic. Therefore, damage cannot be detected early. In contrast, the AE method can be detected at an early stage, but at present, no specific evaluation method has been established. Moreover, neither the AE method nor the vibration method has established a remaining life estimation technique.

  In the measurement of the conventional AE method, a threshold is set, and an AE signal waveform having an amplitude value exceeding the threshold is measured. FIG. 13 shows a standard adopted in a method for diagnosing the bearing life from the change in the amplitude of the AE signal waveform using the conventional AE method announced by the inventor. The amplitude of the AE signal waveform changes at about 50% or about 20% of the remaining life and reaches final destruction. The initial stage is an initial stage of fatigue of the sliding surface of the bearing, and a structural change and microscopic damage occur. A relatively large damage occurs after the remaining life of 20%. If this AE change can be evaluated, a rough remaining life can be evaluated. Conventionally, however, the number of AE signals generated and many AE parameters change due to a threshold setting change for determining a change in amplitude. Therefore, the conventional method has a problem that the diagnosis result depends on the operator.

  An object of the present invention is to provide a bearing diagnosis method and system that eliminates the need for setting a threshold and that can simplify setting of conditions during measurement.

  The present invention is directed to a bearing diagnosis method for diagnosing the life of a bearing in a rotary machine using the AE method.

  In the basic method of the present invention, AE waveform data of a predetermined time length output from an AE sensor mounted in the vicinity of a bearing of a normal rotating machine is s (where s is an integer of 2 or more) at a predetermined time interval. ) Divide and obtain s reference divided waveform data. Here, as the AE sensor, an AE sensor that covers a frequency range of 100 kHz to 500 kHz is used. The predetermined time interval is preferably a time interval in which the shaft of the target bearing makes one rotation or more.

  Next, in the method of the present invention, frequency analysis is performed on s pieces of reference divided waveform data to obtain s pieces of reference frequency analysis data. In the reference frequency analysis data, m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to the occurrence of damage are determined. The method of determining the m frequency bands is determined by a prior test according to the AE signal to be used and the diagnosis target. Since at least one frequency band in which the amplitude value becomes large appears even if the diagnosis target is different, m is set to 1 or more in the present invention. Of course, the diagnosis accuracy can be improved by increasing the number of m rather than decreasing the number of m.

  Next, n reference maximum amplitude values in m frequency bands are obtained from s reference divided waveform data, and an average reference maximum amplitude value of s reference maximum amplitude values in m frequency bands is expressed as m. Ask for one. The above processing is for acquiring basic data for executing the following diagnostic processing.

  Next, an AE sensor is mounted in the vicinity of the bearing of the rotating machine to be diagnosed, and AE waveform data of a predetermined time length output from the AE sensor is divided into n (where n is an integer of 2 or more) at predetermined time intervals. Then, n pieces of divided waveform data for diagnosis are acquired. Then, frequency analysis is performed on the n pieces of diagnostic waveform data to obtain n pieces of diagnostic frequency analysis data. The maximum amplitude values in m frequency bands are obtained from the n diagnostic frequency analysis data, and n × m maximum amplitude values are obtained.

  Next, for each of the m frequency bands, the n maximum amplitude values in the corresponding frequency band are divided by the average reference maximum amplitude value in the corresponding frequency band, and are respectively relativized. Find the maximum amplitude ratio. Then, the existence ratio of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n pieces of divided waveform data for diagnosis belongs is obtained, and the life of the bearing of the rotating machine to be diagnosed is calculated based on the existence ratio. Determine.

  It is known that the maximum amplitude value of the AE waveform data increases as the degree of damage generated in the bearing progresses. However, it is impossible to know the degree of damage only by observing the maximum amplitude value of the AE waveform data. That is, it is not possible to know whether the bearing will be destroyed in a relatively short time or whether there is a period of time until the bearing is destroyed only by observing the maximum amplitude value of the AE waveform data. The inventor determines between the degree of damage that occurs in the bearing and the existence rate of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic divided waveform data belongs. I found that there is a correlation. The present invention is based on this discovery. In the present invention, since diagnosis is performed based on the relative maximum amplitude ratio, it is not necessary to prepare a threshold for diagnosis that is absolutely necessary in the acoustic emission technique (AE technique). Therefore, even if the data varies, the accuracy of diagnosis is not greatly affected. According to the present invention, the degree of bearing damage (or the remaining period until the end of the service life) can be diagnosed by confirming the above-mentioned correlation by a prior test.

  In the present invention, it goes without saying that the knowledge that “the maximum amplitude value of the AE waveform data increases as the degree of damage occurring in the bearing advances” may be used together. In that case, in the pre-processing in the above method, s reference maximum amplitude values are acquired from the s reference divided waveform data, and the average reference maximum amplitude value is obtained from the s reference maximum amplitude values. Further, after AE waveform data having a predetermined time length is divided into n at predetermined time intervals from the vicinity of the bearing of the rotating machine to be diagnosed to obtain n diagnostic divided waveform data, n diagnostic divided waveforms are obtained. N maximum amplitude values are obtained from the data. Then, n relative maximum actual amplitude ratios are obtained by dividing the n maximum amplitude values by the average reference maximum amplitude value, and n relative maximum actual amplitude ratios are p (where p is an integer of 3 or more) types of amplitudes. Replace with level. In addition, the pattern of the combination pattern of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic analysis waveform data belongs and the amplitude level of the corresponding diagnostic divided waveform data Find the presence rate. And based on this pattern presence rate, the lifetime of the bearing of the rotating machine to be diagnosed is determined. When diagnosis is performed based on the pattern presence rate in consideration of the amplitude level as described above, the diagnosis accuracy can be improved as compared with the case where the amplitude level is not considered. The combination pattern is determined by conducting a test in advance. The correlation between the presence rate of the combination pattern and the degree of damage to the bearing may be confirmed in advance by a test.

The probability that the diagnostic frequency analysis data includes a pulse with a characteristic frequency generated due to damage of a specific part of the bearing may be defined as the damage presence rate. Regarding this characteristic frequency, data is published for each bearing manufactured and sold by the bearing manufacturer. For example, when the bearing is a rolling bearing, data on the characteristic frequency of pulses generated when the inner ring is damaged, the outer ring is damaged, or the rolling element is damaged is published. Yes. Inventor studies have shown that the pattern abundance is greater than the damage abundance, which is one proof that the bearing is very close to failure. Therefore, by comparing the magnitude relationship between the pattern presence rate and the damage presence rate and diagnosing the remaining life based on the pattern presence rate, the diagnostic accuracy can be improved as compared with the case of diagnosing only with the pattern presence rate.

  A bearing diagnosis system as a basis of the present invention includes a reference divided waveform data acquisition unit, a reference frequency analysis data acquisition unit, a reference maximum amplitude value acquisition unit, a reference average value calculation unit, a diagnostic divided waveform data acquisition unit, The diagnostic frequency analysis data acquisition unit, the maximum amplitude value acquisition unit, the relative maximum amplitude ratio calculation unit, the existence rate calculation unit, and the life determination unit are provided. In the more specific bearing diagnosis system of the present invention, a reference maximum amplitude value acquisition unit, an average reference maximum amplitude value calculation unit, a diagnostic maximum amplitude value acquisition unit, a diagnostic relative maximum value amplitude ratio calculation unit, A pattern presence rate calculation unit is provided instead of the amplitude level determination unit and the presence rate calculation unit.

  The reference divided waveform data acquisition unit outputs AE waveform data of a predetermined time length output from an AE sensor mounted in the vicinity of a bearing of a normal rotating machine at predetermined time intervals (where s is an integer of 2 or more). Divide and obtain s reference divided waveform data. The reference frequency analysis data acquisition unit performs frequency analysis on s pieces of reference divided waveform data, and acquires s pieces of reference frequency analysis data. In the reference frequency analysis data, the reference maximum amplitude value acquisition unit determines m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to the occurrence of damage, and then s number of frequency bands. S reference maximum amplitude values in m frequency bands are obtained from the reference frequency analysis data. The reference average value calculation unit calculates m reference average values of s reference maximum amplitude values in m frequency bands. The diagnostic divided waveform data acquisition unit outputs AE waveform data of a predetermined time length output from an AE sensor mounted in the vicinity of the bearing of the rotating machine to be diagnosed at a predetermined time interval (where n is 2 or more). Divide by (integer) and obtain n pieces of divided waveform data for diagnosis. The diagnostic frequency analysis data acquisition unit performs frequency analysis on the n pieces of diagnostic divided waveform data and acquires n pieces of diagnostic frequency analysis data. The maximum amplitude value acquisition unit acquires maximum amplitude values in m frequency bands from n diagnostic frequency analysis data, and obtains n × m maximum amplitude values. The relative maximum amplitude ratio calculating unit divides the n maximum amplitude values of the corresponding frequency band by the reference average value in the corresponding frequency band for each of m frequency bands, and relativizes each of the m × n pieces. The relative maximum amplitude ratio is calculated. The abundance ratio calculation unit calculates the abundance ratio of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n pieces of divided waveform data for diagnosis belongs. The life determination unit determines the life of the bearing of the rotating machine to be diagnosed based on the pattern presence rate.

  A reference maximum amplitude value acquisition unit provided in a more specific diagnosis system acquires s reference maximum amplitude values from s reference divided waveform data, and an average reference maximum amplitude value calculation unit calculates s reference maximum amplitude values. From this, the average reference maximum amplitude value is obtained. The maximum amplitude value acquisition unit acquires n maximum amplitude values from the n diagnostic divided waveform data, and the average maximum amplitude value calculation unit calculates an average maximum amplitude value from the n maximum amplitude values. The amplitude level determination unit obtains n relative maximum amplitude ratios by dividing the n maximum amplitude values by the average reference maximum amplitude value, and sets the n relative maximum amplitude ratios to p (where p is an integer of 3 or more). ) Replace with different amplitude levels. Then, the pattern existence ratio calculation unit calculates the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic waveform data for diagnosis belongs and the amplitude level of the corresponding diagnostic divided waveform data. The pattern presence rate of the combination pattern is calculated. The life determination unit determines the life of the bearing of the rotating machine to be diagnosed based on the pattern presence rate.

The diagnostic frequency analysis data further includes a damage presence rate calculation unit that calculates a damage presence rate based on a content rate that includes a pulse of a characteristic frequency that occurs due to damage of a specific part of the bearing. It may be. When providing the damage presence rate calculation unit, it is preferable that the life determination unit diagnoses the remaining life based on the pattern presence rate by comparing the magnitude relationship between the pattern presence rate and the damage presence rate.

It is a block diagram which shows the structure of the bearing diagnostic system of 1st Embodiment. (A) shows an example of AE waveform data (measurement waveform) of a predetermined time length of the filtered AE sensor, and (B) shows the AE waveform data W of FIG. It is a figure which shows an example of the reference | standard division | segmentation waveform data obtained by dividing | segmenting by. (A) shows an example of reference divided waveform data when the bearing is normal, and (B) shows an analysis of the reference divided waveform data of FIG. 3 (A) using a known frequency analysis technique such as Fourier transform. It is a figure which shows an example of the reference frequency analysis data obtained in this way. (A) And (B) is a figure which performs a test in advance and shows data. It is a table | surface used in order to demonstrate the calculation in a reference | standard average value calculating part, and the data memorize | stored in the built-in memory part. It is a table | surface used in order to demonstrate the data memorize | stored in the memory part in a maximum amplitude value acquisition part. It is a table | surface which shows an example of the calculation result of the relative maximum amplitude obtained by calculation. It is a figure used in order to explain the example which digitizes a frequency band. It is a table | surface which shows the example of a damage evaluation standard. It is a table | surface which shows an example of the characteristic frequency data of the bearing which a bearing maker provides to a user as data. It is a table | surface which shows an example of evaluation criteria. It is a block diagram which shows the structure of 2nd Embodiment of the bearing diagnostic system of this invention. It is the figure which showed the standard employ | adopted with the method of diagnosing the lifetime of a bearing from the change of the amplitude of an AE signal waveform using the conventional AE method.

  Embodiments of a bearing diagnosis system and method according to the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a bearing diagnosis system 1 according to the first embodiment. The calculation part and the determination part in this embodiment are mainly realized by using a computer. This bearing diagnosis system includes an AE sensor 3, a reference divided waveform data acquisition unit 5, a divided waveform reference maximum amplitude value acquisition unit 7, an average reference maximum amplitude value calculation unit 9, a reference frequency analysis data acquisition unit 11, Reference maximum amplitude value acquisition unit 13, reference average value calculation unit 15, diagnostic divided waveform data acquisition unit 17, diagnostic frequency analysis data acquisition unit 19, maximum amplitude value acquisition unit 21, and relative maximum amplitude ratio calculation Unit 23, pattern presence rate calculation unit 25, life determination unit 27, diagnostic maximum amplitude value acquisition unit 29, diagnostic relative maximum amplitude ratio calculation unit 31, amplitude level determination unit 33, damage presence rate calculation A unit 35 and a display unit 39 are provided.

  The reference divided waveform data acquisition unit 5 outputs AE waveform data of a predetermined time length output from the AE sensor 3 mounted in the vicinity of a bearing of a normal rotating machine NB at a predetermined time interval (for example, 1 msec) (however, s is an integer of 2 or more), and s pieces of reference divided waveform data are obtained. As the AE sensor 3, an AE sensor that covers a frequency range of 100 kHz to 500 kHz is used. When diagnosing a plurality of bearings used in the motor, first, the AE sensor 3 is mounted on the outer case of the motor located near the installation position of each bearing while the motor is new. At this time, it is preferable to mount the AE sensor 3 at a position that is not affected by other bearings as much as possible. In the present embodiment, a filter for filtering the output of the AE sensor 3 is incorporated in the reference divided waveform data acquisition unit 5 in order to eliminate as much as possible the influence of noise and the presence of other bearings.

  FIG. 2A shows an example of an output waveform of a predetermined time length of the filtered AE sensor 3, that is, AE waveform data (measurement waveform) W. FIG. FIG. 2B shows an example of the reference divided waveform data DW obtained by dividing the AE waveform data W having a predetermined time length by s (where s is an integer of 2 or more) at a predetermined time interval T. ing. In the present embodiment, the predetermined time interval T is set to a time interval that is approximately 1.5 times the time interval in which the shaft of the bearing B to be diagnosed rotates once. When obtaining basic data for diagnosis and when making a diagnosis, the motor is rotated at a predetermined constant speed.

  The reference frequency analysis data acquisition unit 11 performs frequency analysis on the s pieces of reference divided waveform data DW and acquires s pieces of reference frequency analysis data FW. FIG. 3A shows an example of the reference divided waveform data DW when the bearing is normal, and FIG. 3B shows a known frequency such as Fourier transform for the reference divided waveform data DW of FIG. 3A. An example of reference frequency analysis data FW obtained by analysis using an analysis technique is shown.

  The reference maximum amplitude value acquisition unit 13 determines m frequency bands (where m is an integer equal to or greater than 1) frequency bands in which the amplitude value increases due to the occurrence of damage in the reference frequency analysis data FW. S reference maximum amplitude values in m frequency bands are obtained from the reference frequency analysis data FW. Data of m frequency bands can be obtained from the reference frequency analysis data FW using a bandpass filter. In the case where the bearing includes a rolling bearing, it is known as a result of the test that m = 4 is preferable. Therefore, in the following explanation, it is assumed that m = 4 except when the general theory is explained. The method of determining the m frequency bands is determined according to the AE signal to be used and the structure of the bearing to be diagnosed. In practice, a test is performed in advance on a motor having a bearing that is damaged by the same bearing as the bearing to be diagnosed, and data DW and FW as shown in FIGS. 4A and 4B are obtained. It will be determined. Incidentally, in the example of FIG. 3B, the four frequency bands FB1 to FB4 have predetermined bandwidths with 125 kHz, 200 kHz, 300 kHz, and 450 kHz as center frequencies, respectively. In the example of FIG. 3B, the four frequency bands FB1 to FB4 are set to be continuous. However, when a plurality of bands having large amplitude values are separated, the plurality of frequency bands are continuous. There is no need to let them. The reference maximum amplitude value acquisition unit 13 acquires the maximum amplitude value in each frequency band FB1 to FB4 as the reference maximum amplitude value. Specifically, the envelope of the waveform of the reference frequency analysis data is obtained, and the maximum value of each envelope in the four frequency bands FB1 to FB4 is determined as the reference maximum amplitude value. This process is realized using a known envelope setting technique and voltage measurement technique. The processing results are sequentially stored in a memory built in the reference average value calculation unit 15.

  The reference average value calculation unit 15 calculates m reference average values of the s reference maximum amplitude values in the four (m) frequency bands FB1 to FB4. FIG. 5 is a table used for explaining the calculation in the reference average value calculation unit 15 and the data stored in the built-in memory unit. In FIG. 5, data No. is a data number assigned to s pieces of reference frequency analysis data FW. The “band 1” to “band 4” columns mean the frequency bands FB1 to FB4. The “whole area” column means the whole area of the reference frequency analysis data FW described later. Reference maximum amplitude values a1 to as4 are recorded in the “band 1” to “band 4” column and the “whole area” column. The reference average value calculation unit 15 calculates the reference average values Ave1 to Ave1 to s of the reference maximum amplitude values in the four (m) frequency bands FB1 to FB4, and stores the results in a built-in memory unit. .

  The divided waveform reference maximum amplitude value acquisition unit 7 acquires s reference maximum amplitude values from the entire area of the s reference divided waveform data DW. The reference maximum amplitude value can also be obtained by the same method as the above-described reference maximum amplitude value. In the “whole area” column shown in FIG. 5, reference maximum amplitude values a1 to as acquired from each reference divided waveform data DW are described. The reference maximum amplitude values a 1 to as are stored in a memory unit built in the average reference maximum amplitude value calculation unit 9. The average reference maximum amplitude value calculator 9 calculates an average value of the s reference maximum amplitude values a1 to as and stores the calculation result as an average reference maximum amplitude value Ave0 in a built-in memory unit.

  The diagnostic divided waveform data acquisition unit 17 outputs AE waveform data of a predetermined time length output from the AE sensor 3 mounted in the vicinity of the bearing B of the rotating machine to be diagnosed at predetermined time intervals (where n is N pieces of divided waveform data DW ′ are obtained. FIG. 4A shows an example of the reference divided waveform data DW obtained from the AE signal data measured from the damaged bearing, and FIG. 4B shows the reference divided waveform of FIG. An example of the reference frequency analysis data FW obtained by analyzing the data DW using a known frequency analysis technique such as Fourier transform is shown.

  The diagnostic frequency analysis data acquisition unit 19 performs frequency analysis on the n pieces of diagnostic divided waveform data DW ′ and acquires n pieces of diagnostic frequency analysis data FW ′. The operation of the diagnostic frequency analysis data acquisition unit 19 is the same as the operation of the reference frequency analysis data acquisition unit 11 described above.

  The maximum amplitude value acquisition unit 21 acquires maximum amplitude values in four (m) frequency bands FB1 to FB4 from n diagnostic frequency analysis data FW ′, and obtains n × m maximum amplitude values. Ask. The method for obtaining the maximum amplitude value in the maximum amplitude value acquisition unit 21 is the same as the method for obtaining the maximum amplitude value in the reference maximum amplitude value acquisition unit 13 described above. The n × m maximum amplitude values acquired by the maximum amplitude value acquisition unit 21 are stored in a memory unit built in the maximum amplitude value acquisition unit 21. FIG. 6 is a table used to explain data stored in the memory unit in the maximum amplitude value acquisition unit 21. In FIG. 6, data No. is a data number assigned to n pieces of diagnostic frequency analysis data FW ′. The “band 1” to “band 4” columns mean the frequency bands FB1 to FB4. The “whole area” column means the whole area of diagnostic frequency analysis data FW ′ described later. Reference maximum amplitude values b11 to bn4 are recorded in the "Band 1" to "Band 4" and "Whole area" columns.

  As shown in the table of FIG. 7, the relative maximum amplitude ratio calculation unit 23 calculates n maximum amplitude values of the corresponding frequency band for each of the four (m) frequency bands FB1 to FB4 in the corresponding frequency band. Dividing by the reference average values Ave1 to Ave4 (for example, b11 / Ave1 to bn4 / Ave4) and relativizing them, the m × n relative maximum amplitude ratios A1 to An4 are calculated. The calculation result is stored in a memory unit built in the relative maximum amplitude ratio calculation unit 23.

  As shown in the table of FIG. 6, the diagnostic maximum amplitude value acquisition unit 29 acquires n maximum amplitude values b1 to bn from n reference divided waveform data DW 'and stores them in a built-in memory unit. . As shown in the table of FIG. 7, the diagnostic relative maximum amplitude ratio calculation unit 31 divides the n maximum amplitude values b1 to bn by the average reference maximum amplitude value Ave0 (for example, b1 / Ave0 to bn / Ave0). , N relative maximum amplitude ratios A1 to An are obtained.

  The amplitude level determination unit 33 replaces the n relative maximum amplitude ratios A1 to An with p (where p is an integer of 3 or more) types of amplitude levels. In the present embodiment, as shown in FIG. 8, when the relative maximum amplitude ratios A1 to An correspond to 60 dB or less, the level 1 is set, and when the relative maximum amplitude ratios A1 to An correspond to 60 to 70 dB, the level 2 is set and 70 to 80 dB. The amplitude level determination unit 33 performs the replacement to level 3 when it corresponds to, level 4 when it corresponds to 80 to 90 dB, and level 5 when 90 dB or more.

  Then, the pattern existence ratio calculation unit 25 selects the frequency bands FB1 to FB4 to which the maximum value among m relative maximum amplitude ratios (for example, A11 to A14) corresponding to each of the n pieces of diagnostic waveform data FW ′ for diagnosis belongs. Digitize. As shown in FIG. 8, the frequency band FB1 is digitized as 1, the frequency band FB2 is digitized as 2, the frequency band FB3 is numeric as 3, and the frequency band FB4 is numeric as 4. Next, the pattern existence ratio calculation unit 25 determines a combination pattern with the amplitude level (1 to 5) of the corresponding diagnostic divided waveform data DW ′. In this embodiment, a numerical value indicating the amplitude level is placed at the 10's place, and a numerical value showing the frequency band is placed at the 1's place, and the combination pattern is specified by the numerical value. Specifically, as shown in FIG. 8, if the amplitude level is level 1 and the frequency band including the highest amplitude is FB1 = 1, the combination pattern is expressed as 11. In the case of the combination pattern specified as “23”, the amplitude level is level 2, and the frequency band including the highest amplitude is FB3 = 3. In this way, the pattern presence rate calculation unit 25 determines the combination pattern of the n pieces of diagnostic analysis waveform data FW ′ (data Nos. 1 to n) as shown in the table shown in the central part of FIG. As shown in FIG. 8, the pattern presence rate calculation unit 25 calculates the presence rate of the combination pattern with the largest number of data by counting the number of data of the combination pattern. For example, as can be seen from the tabulation result shown in FIG. 8, if the number of combination patterns 13 is the largest, the pattern presence rate of the combination pattern 13 is 5 / n.

  When the life determination unit 27 determines the life of the bearing based on the pattern presence rate, a damage degree evaluation standard as shown in FIG. 9 is prepared in advance. The lifetime is determined according to which of the evaluation patterns in FIG. 9 corresponds to the combination pattern in which the pattern presence rate calculated by the pattern presence rate calculation unit 25 is the largest. That is, assuming that the combination pattern having the largest pattern presence rate is “34”, the remaining lifetime is determined to be 50% to 20% according to the evaluation criteria of FIG. In FIG. 9, the notation [1 *] means that any numerical value of 1 to 4 may be entered as “*”. This evaluation shows that if the amplitude level is low regardless of the frequency band, the damage is small and the lifetime is still sufficient. And the notation that the combination pattern is [4 *] or [5 *] also means that any numerical value of 1 to 4 may be entered as “*”. This evaluation shows that if the amplitude level is as high as 4 or 5 regardless of the frequency band, the damage is high and the lifetime is short. In the case of [21] to [23] or [31] to [33], there is damage, but the remaining life is not yet shorter than 50%. The evaluation criteria shown in FIG. 9 are determined in advance according to the type of bearing.

  Since damage to the bearing occurs at a period of a characteristic frequency determined by the structure and the number of rotations of the bearing, a “damage existence rate” considering a specific frequency can be introduced into the life determination. In the present embodiment, the life determination unit 27 further increases the determination accuracy by considering the damage presence rate. Therefore, in the present embodiment, a damage presence rate calculation unit 35 is provided. The damage existence rate calculation unit 35 includes a pulse of a characteristic frequency generated due to damage of a specific part of the bearing in the n pieces of diagnostic frequency analysis data FW ′ acquired from the diagnostic frequency analysis data acquisition unit 19. The probability of inclusion is calculated as the damage presence rate. FIG. 10 is an example of bearing characteristic frequency data provided to the user as a document by the bearing manufacturer. Here, the “damage abundance rate” indicates the probability that a pulse waveform associated with a scratch on the slowest part of the bearing is included in the divided waveform when the bearing is a rolling bearing. (Wavelength length and number of divisions). For example, in the characteristic frequency example of FIG. 10, consider a case where the waveform length (predetermined time length) of the AE waveform data is set to 30 msec at 3000 rpm with respect to the bearing. One rotation of 3000 rpm is 20 msec, and the waveform of the measured AE waveform data has a waveform length of 1.5 rotations. This is the slowest rolling element in the pulse generation cycle of the inner ring, outer ring, and rolling element of the bearing, and the generation period (characteristic frequency) is 249 Hz = about 4 msec. Therefore, there is a possibility that a pulse generated due to the rolling element damage appears 7.4 times in the waveform of 30 msec. When the 30msec length AE waveform data is divided into 30 for 1msec length and the divided waveform data for reference and diagnosis (DW, DW ') is obtained, the "damage presence rate" is 7.4 / 30 = about 25% . The damage presence rate calculation unit 35 performs the above calculation. Instead of providing the damage presence rate calculation unit 35 as in the present embodiment, the damage presence rate may be stored in the life determination unit 27 as a constant.

When the life determination unit 27 uses the damage presence rate, for example, the evaluation criteria shown in FIG. 11 can be used. As shown in FIG. 11, in the “remaining life prediction”, whether the measured time is any of [remaining life 50% or more], [remaining life 50% to 20%], or [remaining life 20% or less]. Judging. In “damage sign detection”, it is determined whether there is any minor damage that requires continuous observation. Therefore, in this evaluation standard, when the pattern presence rate is A and the damage presence rate is B, the magnitude relationship between the pattern presence rate A and the damage presence rate B is considered as an evaluation factor. That is, according to this evaluation standard, even if the evaluation pattern is the same combination pattern [24], [34], if the pattern existence ratio A <damage existence ratio B, it is determined that the remaining life is 50% or more. In the case of pattern presence rate A ≧ damage presence rate B, it is determined that the remaining life is 50% to 20%. In the criteria of FIG. 11, when the evaluation pattern is other than [24] and [34], the magnitude relationship between the pattern presence rate A and the damage presence rate B is not considered at all. This is because it is possible to accurately diagnose only with the evaluation pattern (combination pattern) without using the magnitude relationship between the pattern presence rate A and the damage presence rate B. The diagnosis result is displayed on the display unit 39 by characters or the like.

  The relationship between the pattern abundance ratio A and the damage abundance ratio B is not limited to the example of FIG. 11, and depending on the specifications of the bearing, a result obtained by multiplying the damage abundance ratio B by a correction coefficient may be used. A result obtained by adding / subtracting the correction value to / from the existence ratio B may be used.

  According to the present embodiment, it is ranked (contribution rank pattern) which frequency band contributes more to the change in the amplitude value of the AE signal for the four frequency bands. As a result, it is possible to characterize the existence of each damage stage, evaluate which damage stage it belongs to, that is, predict the remaining life, and determine the cause of damage. Further, since the determination does not depend on the absolute value, it is possible to reduce the influence of sensitivity correction, threshold setting, and the like that are necessary for the conventional AE measurement method. In the present embodiment, it is possible to detect damage (with damage) that requires continuous observation, although the remaining life cannot be predicted. Furthermore, since it is difficult to be influenced by the size of the bearing and the above sensitivity, the evaluation can be performed regardless of the operating conditions and the bearing type of the target device. In this embodiment, a waveform having a certain length is divided, and a waveform having a certain time is collected without setting a threshold value. Therefore, measurement can be easily performed.

  In the embodiment of FIG. 1, the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic waveform data for diagnosis belongs, and the amplitude level of the corresponding divided waveform data for diagnosis. A combination pattern is determined based on the combination of and the pattern presence rate of the combination pattern. However, without using the amplitude level, the frequency band FB1 to FB4 to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic waveform data for diagnosis belongs is set to four patterns, and the pattern existence rate is calculated. It may be determined. FIG. 12 is a block diagram showing the configuration of the second embodiment of the bearing diagnosis system of the present invention used in this case. In the embodiment of FIG. 12, the same reference numerals as those of the embodiment of FIG. 1 are added with the number of 100 added to the number of reference numerals attached to FIG. 1. In the embodiment of FIG. 12, the reference average value calculation unit 115 calculates the reference average values Ave1 to Ave4 of FIG. 5, and the maximum amplitude value acquisition unit 121 calculates the 4 × n maximum amplitude values b11 to bn4 of FIG. The relative maximum amplitude ratio calculation unit 123 acquires the 4 × n relative maximum amplitude ratios A11 to An4 in FIG. The abundance ratio calculation unit 125 digitizes the frequency band for the relative maximum amplitude ratios A11 to An4, digitizes and patterns the relative maximum amplitude ratios A11 to An4, and aggregates the digitized patterns. That is, the number of data patterned into numerical values of 1 to 4 is totaled, and the pattern presence rate is calculated. Assuming that the total numbers of the numerical values 1 to 4 are X1 to X4, X1 / n to X4 / n are pattern existence ratios. The life determination unit 127 diagnoses the life of the bearing based on the order of frequency bands in which the pattern presence rate increases. For example, when the existence ratio of the frequency band FB1 is the largest, it is determined as normal, and when both of the existence ratios of the frequency bands FB1 and FB4 are large, a tissue change or minute damage occurs in the rolling elements of the bearing. If the existence ratios of the frequency bands FB1, FB2, and FB3 are both increased, it is determined that macroscopic damage and cracks have occurred in the rolling elements of the bearing. In this embodiment, although the diagnostic accuracy is lower than that in the embodiment of FIG. 1, far more objective diagnosis is possible than in the case of simply comparing the amplitude value and the threshold value. Note that the criterion used in the life determination unit 127 in the present embodiment is not limited to the reference content described above, and may be appropriately determined by a preliminary test.

Also, in the embodiment of FIG. 12, as in the embodiment of FIG. 1, it is needless to say that the determination may be made by comparing the magnitude relationship between the pattern presence rate and the damage presence rate. For example, even when the pattern existence ratios of the frequency bands FB1 and FB4 both increase, if the largest pattern existence ratio <damage existence ratio, it is determined that the remaining life is 50% or more, and the frequency bands FB1 and FB4 Even if the abundance ratios of both are not increased, if the largest pattern abundance ratio A ≧ damage abundance ratio B, it can be determined that the remaining lifetime is 50% to 20%. In addition, when comparing the magnitude relationship between the pattern presence rate and the damage presence rate, the determination standard is determined by actually conducting a test in advance, so that the determination criterion of the present embodiment is not limited.

  According to the present invention, since diagnosis is performed based on the relative maximum amplitude ratio, it is not necessary to prepare a threshold value for diagnosis that is absolutely necessary in the AE technique. Therefore, even if there is variation in data, there is an advantage that the accuracy of diagnosis is not greatly affected.

1 Bearing Diagnosis System 3 AE Sensor 5 Reference Divided Waveform Data Acquisition Unit 7 Divided Waveform Reference Maximum Amplitude Value Acquisition Unit 9 Average Reference Maximum Amplitude Value Calculation Unit 11 Reference Frequency Analysis Data Acquisition Unit 13 Reference Maximum Amplitude Value Acquisition Unit 15 Reference Average Value Calculation Unit 17 Diagnosis divided waveform data acquisition unit 19 Diagnosis frequency analysis data acquisition unit 21 Maximum amplitude value acquisition unit 23 Relative maximum amplitude ratio calculation unit 25 Pattern existence ratio calculation unit 27 Life determination unit 29 Diagnosis maximum amplitude value acquisition unit 31 Diagnosis Relative maximum amplitude ratio calculation unit 33 Amplitude level determination unit 35 Damage presence rate calculation unit 39 Display unit

Claims (6)

A bearing diagnosis method for diagnosing the life of a bearing in a rotating machine using acoustic emission,
The AE waveform data of a predetermined time length output from the AE sensor mounted in the vicinity of the bearing of the normal rotating machine is divided into s (where s is an integer of 2 or more) at predetermined time intervals, and s reference values are obtained. Acquire divided waveform data,
Obtaining s reference maximum amplitude values from the s reference divided waveform data;
An average reference maximum amplitude value is obtained from the s reference maximum amplitude values,
Frequency-analyzing the s reference divided waveform data to obtain s reference frequency analysis data;
In the reference frequency analysis data, m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to occurrence of damage are defined,
Obtaining s reference maximum amplitude values in the m frequency bands, respectively, from the s reference frequency analysis data;
Obtaining m average reference maximum amplitude values of the s reference maximum amplitude values in the m frequency bands;
The AE sensor is mounted in the vicinity of the bearing of the rotating machine to be diagnosed, and the AE waveform data of the predetermined time length output from the AE sensor is n at the predetermined time interval (where n is an integer of 2 or more) ) Divide and obtain n diagnostic waveform data,
Obtaining n maximum amplitude values from the n divided waveform data for diagnosis,
Dividing the n maximum amplitude values by the average reference maximum amplitude value to obtain n relative maximum amplitude ratios;
The n relative maximum amplitude ratios are replaced with p (where p is an integer of 3 or more) kinds of amplitude levels,
Frequency analysis of the n diagnostic divided waveform data to obtain n diagnostic frequency analysis data;
Obtaining the maximum amplitude value in each of the m frequency bands from the n diagnostic frequency analysis data to obtain n × m maximum amplitude values;
For each of the m frequency bands, m × n relative maximums are obtained by dividing the n maximum amplitude values of the corresponding frequency band by the reference average value in the corresponding frequency band and relativizing each. Find the amplitude ratio,
A combination pattern of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic waveform data for diagnosis belongs and the amplitude level of the corresponding divided waveform data for diagnosis Find the pattern presence rate,
A bearing diagnosis method comprising determining a life of a bearing of the rotating machine to be diagnosed based on the pattern presence rate. In the diagnostic frequency analysis data, the probability of including a pulse of a characteristic frequency that occurs due to damage of a specific part of the bearing is defined as a damage presence rate,
2. The bearing diagnosis method according to claim 1, wherein the remaining life is diagnosed based on the pattern existence ratio by comparing a magnitude relationship between the pattern existence ratio and the damage existence ratio. A bearing diagnosis method for diagnosing the life of a bearing in a rotating machine using acoustic emission,
The AE waveform data of a predetermined time length output from the AE sensor mounted in the vicinity of the bearing of the normal rotating machine is divided into s (where s is an integer of 2 or more) at predetermined time intervals, and s reference values are obtained. Acquire divided waveform data,
Frequency-analyzing the s reference divided waveform data to obtain s reference frequency analysis data;
In the reference frequency analysis data, m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to the occurrence of damage are defined,
Obtaining s reference maximum amplitude values in the m frequency bands, respectively, from the s reference divided waveform data;
Obtaining m reference average values of s reference maximum amplitude values in the m frequency bands;
The AE sensor is mounted in the vicinity of the bearing of the rotating machine to be diagnosed, and the AE waveform data of the predetermined time length output from the AE sensor is n at the predetermined time interval (where n is an integer of 2 or more) ) Divide and obtain n diagnostic waveform data,
Frequency analysis of the n diagnostic divided waveform data to obtain n diagnostic frequency analysis data;
Obtaining the maximum amplitude value in each of the m frequency bands from the n diagnostic frequency analysis data to obtain n × m maximum amplitude values;
For each of the m frequency bands, m × n relative maximums are obtained by dividing the n maximum amplitude values of the corresponding frequency band by the reference average value in the corresponding frequency band and relativizing each. Find the amplitude ratio,
Determining the existence rate of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n pieces of diagnostic divided waveform data belongs;
A bearing diagnosis method comprising: determining a life of a bearing of the rotating machine to be diagnosed based on the presence rate. A bearing diagnosis system for diagnosing the life of a bearing in a rotating machine using acoustic emission,
The AE waveform data of a predetermined time length output from the AE sensor mounted in the vicinity of the bearing of the normal rotating machine is divided into s (where s is an integer of 2 or more) at predetermined time intervals, and s reference values are obtained. A reference split waveform data acquisition unit for acquiring split waveform data;
A reference maximum amplitude value acquisition unit for acquiring s reference maximum amplitude values from the s reference divided waveform data;
An average reference maximum amplitude value calculation unit for obtaining an average reference maximum amplitude value from the s reference maximum amplitude values;
A frequency analysis of the s reference divided waveform data to obtain s reference frequency analysis data;
In the reference frequency analysis data, m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to the occurrence of damage are determined, and each of the s reference frequency analysis data is determined. A reference maximum amplitude value acquisition unit for acquiring s reference maximum amplitude values in the m frequency bands;
A reference average value calculation unit that calculates m reference average values of the s maximum amplitude values in the m frequency bands;
The AE waveform data of the predetermined time length output from the AE sensor mounted in the vicinity of the bearing of the rotating machine to be diagnosed is divided into n (where n is an integer of 2 or more) at the predetermined time interval, and n A divided waveform data acquisition unit for diagnosis that acquires the divided waveform data for diagnosis;
A maximum amplitude value acquisition unit for acquiring n maximum amplitude values from the n diagnostic divided waveform data;
A relative maximum amplitude ratio calculating unit that obtains n relative maximum amplitude ratios by dividing the n maximum amplitude values by the average reference maximum amplitude value;
An amplitude level determination unit that replaces the n relative maximum amplitude ratios with p (where p is an integer of 3 or more) types of amplitude levels;
A frequency analysis data acquiring unit for frequency analysis of the n divided waveform data for diagnosis to acquire n pieces of frequency analysis data for diagnosis;
A maximum amplitude value acquisition unit for acquiring maximum amplitude values in the m frequency bands from the n diagnostic frequency analysis data, and obtaining n × m maximum amplitude values;
For each of the m frequency bands, m × n relative maximums are obtained by dividing the n maximum amplitude values of the corresponding frequency band by the reference average value in the corresponding frequency band and relativizing each. A relative maximum amplitude ratio calculator for calculating the amplitude ratio;
A combination pattern of the frequency band to which the maximum value among the m relative maximum amplitude ratios corresponding to each of the n diagnostic waveform data for diagnosis belongs and the amplitude level of the corresponding divided waveform data for diagnosis A pattern presence rate calculator for calculating the pattern presence rate;
A bearing diagnosis system comprising: a life determination unit that determines a life of a bearing of the rotating machine to be diagnosed based on the pattern presence rate. The diagnostic frequency analysis data further includes a damage presence rate calculating unit that calculates a probability of including a pulse of a characteristic frequency generated due to damage of a specific part of the bearing as a damage presence rate,
5. The bearing diagnosis system according to claim 4, wherein the life determination unit compares a relationship between the pattern existence ratio and the damage existence ratio and diagnoses a remaining life based on the pattern existence ratio. A bearing diagnosis system for diagnosing the life of a bearing in a rotating machine using acoustic emission,
The AE waveform data of a predetermined time length output from the AE sensor mounted in the vicinity of the bearing of the normal rotating machine is divided into s (where s is an integer of 2 or more) at predetermined time intervals, and s reference values are obtained. A reference split waveform data acquisition unit for acquiring split waveform data;
A frequency analysis of the s reference divided waveform data to obtain s reference frequency analysis data;
In the reference frequency analysis data, m frequency bands (where m is an integer of 1 or more) whose amplitude value increases due to occurrence of damage are determined, and each of the s reference divided waveform data is determined. A reference maximum amplitude value acquisition unit for acquiring s reference maximum amplitude values in the m frequency bands;
A reference average value calculation unit for calculating m reference average values of the s reference maximum amplitude values in the m frequency bands;
The AE waveform data of the predetermined time length output from the AE sensor mounted near the bearing of the rotating machine to be diagnosed is divided into n at the predetermined time interval to obtain n pieces of divided waveform data for diagnosis. A divided waveform data acquisition unit for diagnosis,
A frequency analysis data acquiring unit for frequency analysis of the n divided waveform data for diagnosis to acquire n pieces of frequency analysis data for diagnosis;
A maximum amplitude value acquisition unit for acquiring maximum amplitude values in the m frequency bands from the n diagnostic frequency analysis data, and calculating n × m maximum amplitude values;
For each of the m frequency bands, m × n relative maximums are obtained by dividing the n maximum amplitude values of the corresponding frequency band by the reference average value in the corresponding frequency band and relativizing each. A relative maximum amplitude ratio calculator for calculating the amplitude ratio;
An abundance ratio calculating unit that calculates an abundance ratio of the frequency band to which a maximum value among the m relative maximum amplitude ratios corresponding to each of the n divided waveform data for diagnosis belongs;
A bearing diagnosis system comprising: a life determination unit that determines the life of a bearing of the rotating machine to be diagnosed based on the presence rate.