JP6507297B1 - Method and apparatus for diagnosing abnormality of rolling bearing, abnormality diagnosis program - Google Patents

Abstract

An object of the present invention is to make it possible to estimate the type of damage to a bearing even when the bearing specifications can not be grasped.
When a measurement start instruction is issued together with a diagnostic rotational frequency, vibrational acceleration is measured and recorded, and rotational frequency and vibrational acceleration are correlated and recorded (S1 to S6). If all measurements have been completed in S7, the rotation synchronization component enhanced waveform is calculated in S8, and in S9 the ratio fundamental frequency and the specific vibration source rotation frequency are assumed, and the ratio fundamental frequency of the rotation synchronization component enhanced waveform X It is determined whether a vibration peak is present at the position of natural number ± specific vibration source rotational frequency, and the ratio (match rate) at which the vibration peak is present is calculated. If it is determined in S10 that the search has been completed in the predetermined search range, then in S11, values of the relative vibration source rotational frequency for the ratio fundamental frequency and the specific vibration source rotational frequency whose coincidence rate is equal to or higher than the predetermined threshold In accordance with, the type of bearing abnormality is determined, and the diagnosis result is displayed in S12.
[Selected figure] Figure 4

Description

  The present invention relates to a method, apparatus, and program for diagnosing an abnormality in a rolling bearing that is used in a machine tool or the like and supports a rotating body such as a main shaft.

When an abnormality such as damage to the inner ring occurs in the rolling bearing that supports the rotating body, vibration occurs. Since the force generated by the abnormality of the bearing is not a simple sine wave, vibrations of harmonic frequency components are simultaneously observed. The frequency (characteristic frequency) of the vibration generated at this time is proportional to the rotation frequency, and can be calculated from the rotation frequency of the rotating body and bearing specifications.
For example, in patent document 1, the waveform which measured vibration acceleration and performed envelope processing, frequency analysis, etc. is displayed with the characteristic frequency of the inner ring, the outer ring, the rolling element, etc. which were calculated based on a predetermined relational expression. A method of identifying a damaged portion (inner ring, outer ring, rolling element, etc.) of a rolling bearing by indicating whether or not there is a corresponding vibration peak is shown.
In Patent Document 2, vibration is measured, envelope processing and frequency analysis are performed, and the values of the feature frequencies of the inner ring, outer ring, and rolling element calculated based on a predetermined relational expression are extracted to obtain damaged portions (inner ring, outer ring , Rolling elements) using different threshold values.

JP-A-63-297813 Patent No. 5146008

  However, in order to use the threshold value according to the identification of the damaged part and the damaged part, bearing specifications for calculating the fundamental frequency and the characteristic frequency are required. For this reason, in the bearing diagnosis system sold by the bearing manufacturer, diagnosis can be made by inputting the type of its own bearing whose bearing manufacturer has grasped the bearing specifications, but bearing specifications of other companies' bearings In general, the implementation requires manual input. In this case, there is a problem that the bearing can not be diagnosed if the specifications of the bearing can not be obtained. In addition, there is a problem that it is difficult even to grasp the type of bearing used in the machine except the machine manufacturer.

  Therefore, the present invention has been made in view of the above problems, and it is possible to estimate the type of damage to the bearing even when the bearing specifications of the bearing used in the machine to be diagnosed can not be grasped. It is an object of the present invention to provide a rolling bearing abnormality diagnosis method, an abnormality diagnosis device, and an abnormality diagnosis program that can be performed.

In order to achieve the above object, the invention according to claim 1 is a method of diagnosing an abnormality in a rolling bearing supporting a rotating body,
A vibration measurement step of measuring the vibration of the rotating body a plurality of times ;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency A rotation synchronization component emphasis waveform calculation step of calculating an average of the ratio frequencies for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
Based on the regularity of the vibration peak position of the rotation synchronization component emphasis waveform , a specific vibration source rotation frequency which is a ratio of a vibration source rotation frequency which is a frequency at which a force generation direction changes in the rolling bearing If the value of the specific vibration source rotational frequency is 1, it is determined that the inner ring of the rolling bearing is damaged, and the value of the specific vibration source rotational frequency is the rotational frequency of the rolling elements of the rolling bearing. In the case of the possible range of the value divided by the frequency, it is judged that the rolling element is damaged, and when the value of the specific vibration source rotational frequency is 0, the bearing damage judged as the outer ring damage of the rolling bearing Performing a type determination step.
The invention according to claim 2 is the configuration according to claim 1 , wherein the estimation of the specific vibration source rotational frequency is calculated on the assumption of a ratio fundamental frequency and a specific vibration source rotational frequency in the rotation synchronization component emphasis waveform. Calculating the match rate, which is the rate at which the vibration peak exists at a specific specific frequency, and if the match rate exceeds a predetermined threshold, the specific vibration source rotational frequency assumed to be the assumed relative fundamental frequency It is determined that the regularity represented by is in the rotation synchronization component emphasis waveform, and the assumed specific vibration source rotation frequency is adopted.
The invention according to claim 3, in the structure according to claim 1 or 2, in the vibration measuring step, and measuring the vibrations at different rotational frequencies.
In order to achieve the above object, the invention according to claim 4 is a method of diagnosing an abnormality in a rolling bearing supporting a rotating body,
A vibration measurement step of measuring the vibration of the rotating body a plurality of times;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency A rotation synchronization component emphasis waveform calculation step of calculating an average of the ratio frequencies for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
The rotation synchronization component enhancement is performed using a machine learning model learned using teacher data in which the rotation synchronization component enhancement waveform is input and at least one of inner ring injury, rolling body injury, and outer ring injury is output. And performing a bearing damage type determination step of inputting a waveform to determine the type of damage to the rolling bearing .
In order to achieve the above object, the invention according to claim 5 is an apparatus for diagnosing an abnormality in a rolling bearing supporting a rotating body,
Vibration measuring means for measuring the vibration of the rotating body a plurality of times ;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency Rotation synchronization component emphasis waveform calculation means for calculating an average of the relative frequency for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
Based on the regularity of the vibration peak position of the rotation synchronization component emphasis waveform , a specific vibration source rotation frequency which is a ratio of a vibration source rotation frequency which is a frequency at which a force generation direction changes in the rolling bearing If the value of the specific vibration source rotational frequency is 1, it is determined that the inner ring of the rolling bearing is damaged, and the value of the specific vibration source rotational frequency is the rotational frequency of the rolling elements of the rolling bearing. In the case of the possible range of the value divided by the frequency, it is judged that the rolling element is damaged, and when the value of the specific vibration source rotational frequency is 0, the bearing damage judged as the outer ring damage of the rolling bearing And a type determination unit.
The invention according to claim 6 is characterized in that, in the configuration according to claim 5 , the bearing damage type discriminating means comprises a relative fundamental frequency corresponding to damage to the rolling bearing , the relative vibration source rotational frequency, and the relative fundamental frequency Based on the value of the specific vibration source rotational frequency, the existence of the vibration peak position for each type of damage of the rolling bearing is confirmed using a specific specific frequency calculated from the specific vibration source rotational frequency To determine the type of the damage ;
It is characterized by including display means for displaying together the relative fundamental frequency and the relative vibration source rotational frequency.
The invention according to claim 7 is characterized in that, in the configuration according to claim 5 , the bearing damage type discriminating means comprises a relative fundamental frequency corresponding to damage of the rolling bearing , the relative vibration source rotational frequency, and the relative fundamental frequency. Based on the value of the specific vibration source rotational frequency, the existence of the vibration peak value for each type of damage of the rolling bearing is confirmed using a specific specific frequency calculated from the specific vibration source rotational frequency To determine the type of the damage ;
It is characterized by comprising display means for displaying together the position of the specific ratio frequency and the rotation synchronization component emphasizing waveform.
In order to achieve the above object, the invention according to claim 8 is an abnormality diagnosis program for a rolling bearing, to a computer to which vibration of a rotating body measured at a predetermined rotational frequency is input together with the rotational frequency, A rotation synchronous component emphasis waveform calculation step and a bearing damage type determination step in the method of diagnosing abnormality of a rolling bearing according to any one of claims 1 to 4 are performed.

According to the present invention, a rotation synchronization component emphasizing waveform for emphasizing a vibration component of a frequency proportional to the rotation frequency of the rotating body is calculated with respect to the measured vibration, and the regularity of the vibration peak position of the rotation synchronization component emphasizing waveform is calculated. Since the type of damage in the rolling bearing is determined, the type of bearing abnormality can be determined from the regularity of the vibration peak without using bearing specifications.
In particular, according to the invention of claim 3 , in addition to the above-mentioned effects, in the vibration measurement step, vibrations are measured at a plurality of different rotational frequencies, so the influence of disturbance is further reduced and the regularity of the vibration peak is captured. Since this becomes easy, the estimation accuracy of the bearing abnormality is improved.
In particular, according to the invention of claim 4 , in addition to the above effects, in the bearing damage type determination step, at least the presence or absence of the inner ring flaw, the presence or absence of the rolling element flaw, and the presence or absence of the outer ring flaw is input using the rotation synchronous component emphasis waveform. Using the machine learning model learned using teacher data with one output as the rotation synchronization component emphasis waveform is input to determine the type of damage to the rolling bearing, the vibration peak is more versatile than the process considered by humans. It is possible to determine the type of bearing abnormality from the regularity.
In particular, according to the sixth aspect of the present invention, in addition to the above effects, by providing display means for displaying together the ratio fundamental frequency and the specific vibration source rotational frequency, using the value of the frequency displayed as a numeral Other bearing diagnosis processes can be performed.
In particular, according to the invention of claim 7 , in addition to the above effect, the regularity of the vibration peak is correctly extracted by providing the display means for displaying together the position of the specific ratio frequency and the rotation synchronization component emphasizing waveform. Since it is possible to visually grasp whether or not it has been done, it is possible to easily verify the validity of the diagnostic result. In addition, the specific frequency to be focused on in each type of damage can be easily understood.

It is a functional block diagram of an abnormality diagnosis device for rolling bearings. It is a rotation synchronous component emphasis waveform when the outer ring and the inner ring are damaged. It is a rotation synchronous component emphasis waveform in case the rolling element is damaged. It is a flowchart of the abnormality diagnosis method. It is a flowchart of the other example of the abnormality-diagnosis method. It is a flowchart of the other example of the abnormality-diagnosis method. This is a screen for displaying a diagnostic rotational frequency and instructing measurement start. It is a screen on which the abnormality diagnosis result by the method of FIG. 4 is displayed. It is a screen on which the abnormality diagnosis result by the method of FIG. 5 is displayed. It is a screen which displays the abnormality diagnosis result by the method of FIG.

Hereinafter, embodiments of the present invention will be described based on the drawings.
FIG. 1 is a functional block diagram showing a configuration in the case where a rolling bearing abnormality diagnosis device is attached to a spindle of a machine tool and used, and will be specifically described based on this drawing.
In the machine tool 100, the spindle 1 is rotatably attached to the spindle housing 2 via a bearing 7 which is a rolling bearing, and a tool 3 for processing is fixed to the spindle 1. The motor 4 drives the spindle 1. The motor 4 is provided with a speed detector 5 so that the measured rotational frequency of the motor 4 is input to the controller 6. The control device 6 controls the current supplied to the motor 4 so as to maintain the rotational frequency of the motor 4 measured by the speed detector 5 at the command rotational frequency input to the control device 6 by the measurer. The display unit 8 displays the current rotational frequency of the motor 4 measured by the speed detector 5.

In the abnormality diagnosis apparatus 200 including a computer, the vibration sensor 9 as a vibration measurement means is attached by magnetic force to a position where the vibration generated by the damage of the bearing 7 of the machine tool 100 can be measured.
The display / operation unit 13 as display means displays the command rotational frequency and the diagnostic result that the measurer should input to the control device 6, and the current rotational frequency of the motor 4 displayed on the display unit 8 is displayed and operated It is possible to input a result of the determination made by the measurer as to whether or not the value displayed in the section 13 matches. The vibration acceleration measured by the vibration sensor 9 is converted into a digital value by the A / D converter 10. Storage unit 11 is from when the measurer determines that the current rotational frequency of motor 4 displayed on display unit 8 matches the value displayed on display / operation unit 13 to when the predetermined data length is reached. The digital value of vibration acceleration is stored together with the rotational frequency at the time of vibration measurement.

  The calculation unit 12 performs multiplication, Fourier transform, calculation of absolute value, and interpolation processing based on the abnormality diagnosis program stored in advance from the rotational frequency and vibration acceleration at the time of vibration measurement stored in the storage unit 11. Rotation synchronization is achieved by calculating a corrected vibration value for a specific frequency (a dimensionless quantity calculated by dividing the frequency by the rotation frequency) and calculating an average for each specific frequency for the corrected vibration value of the rotational frequency at the time of plural vibration measurements. Calculate the component emphasized waveform. Furthermore, the ratio fundamental frequency and the specific vibration source rotational frequency to be described later are estimated from the regularity of the peak position of the rotation synchronization component enhanced waveform, and damage is caused in the inner ring / rolling element / outer ring according to the value of the specific vibration source rotational frequency. Determine where it is happening. That is, the calculation unit 12 functions as a rotation synchronization component emphasis waveform calculation unit and a bearing damage type determination unit.

In the bearing 7, when there is an inner ring flaw, a force at a frequency (basic frequency) at which the rolling element passes the inner ring flaw and a frequency that is a natural number multiple thereof is generated. Since the direction of the inner ring wound rotates with the rotation of the main shaft 1, the direction in which the force is generated changes with the rotation frequency.
In the bearing 7, when there is a rolling element flaw, a force of a frequency (basic frequency) at which the rolling element flaw contacts the inner ring or the outer ring and a frequency that is a natural number multiple thereof is generated. The position of the damaged rolling element changes at a frequency at which the rolling element cage rotates (the rolling element revolves), so the rolling element cage rotates in the direction in which force is generated (the rolling element revolves) It changes with frequency. The rotation of the cage (revolution of the rolling elements) is always slower than the rotation of the main shaft, so the frequency at which the direction in which force is generated changes is greater than 0 Hz and less than the rotation frequency.
In the bearing 7, when there is an outer ring flaw, a force at a frequency (basic frequency) at which the rolling element passes the outer ring flaw and a frequency that is a natural number multiple thereof is generated. Since the direction of the outer ring flaw is constant regardless of the rotation of the main shaft 1, the direction in which the force is generated does not change.
Here, the frequency at which the direction in which the force is generated changes is referred to as the vibration source rotational frequency. If the direction of force generation does not change, the vibration source rotational frequency may be regarded as 0 Hz for convenience.

  By the way, if the direction in which the force is generated changes at the vibration source rotational frequency, looking at the force in a fixed coordinate system, the vibration source rotational frequency will be around the power frequency (natural number multiple of the fundamental frequency). Power of two different frequencies are generated. In a system that can be regarded as linear, vibrational acceleration occurs at the same frequency as the force, so vibrational acceleration is measured in a fixed coordinate system and frequency analysis indicates that if the bearing is damaged, the vibration source rotation is ideal. A pair of vibration peaks is observed at a frequency which is twice as large as the frequency, and a plurality of vibration peaks are observed at such positions that the average frequency of the two vibration peaks is a natural number multiple of the fundamental frequency. By capturing this regularity and estimating the vibration source rotational frequency, it becomes possible to determine the type of abnormality. However, in reality, only by measuring the frequency and analyzing the frequency, the vibration peak of the frequency that is hard to vibrate is buried in the noise, and it is difficult to capture this regularity with certainty.

  The vibration caused by bearing damage synchronized with rotation is always a constant multiple of the rotation frequency, even if it is measured repeatedly or measured at different rotation frequencies. When the dimensionless quantity of the ratio of the frequency of vibration to the frequency of rotation is referred to as the ratio frequency, the vibration caused by the bearing damage synchronized with the rotation has a vibration peak observed at the same ratio frequency regardless of the rotation frequency. become. Therefore, white noise can be reduced by measuring a plurality of times and taking an average, or disturbance components that are not synchronized with white noise and rotation can be reduced by taking an average for each specific frequency by measuring at a plurality of rotational frequencies. By doing this, it is possible to calculate a waveform (rotation synchronization component emphasis waveform) emphasizing the vibration synchronized with the rotation, and it becomes easy to grasp the regularity of the vibration peak caused by the bearing damage.

  When measuring with multiple rotational frequencies, it is better to use the ratio fundamental frequency of the dimensionless quantity obtained by dividing each of the fundamental frequency of vibration and the vibration source rotational frequency by the rotational frequency and the relative vibration source rotational frequency. Cheap. However, since the ratio fundamental frequency takes various values depending on the structure of the bearing, for example, the ratio fundamental frequency of the inner ring flaw of one bearing may be the same magnitude as the ratio fundamental frequency of the outer ring flaw of another bearing. For this reason, it is impossible to determine the type of bearing abnormality even when focusing on the value of the ratio fundamental frequency. However, as described above, depending on the type of bearing damage, the vibration source rotational frequency differs depending on what the rotational frequency is, and no reversal occurs in the magnitude relationship. Therefore, when converted to the specific vibration source rotational frequency, 1 in the case of inner ring damage, and in the case of rolling element damage, values in the range that can be taken of the ratio of rolling element revolution to rotating frequency described later In the case, there are many values around 0.45), and it becomes 0 in the case of the outer ring damage.

  FIG. 2 shows a rotation synchronization component emphasis waveform when the outer ring and the inner ring are damaged, and FIG. 3 shows a rotation synchronization component emphasis waveform when the rolling element is damaged. The regularity of the vibration peak position for each type of bearing damage will be described based on these figures. In addition, in FIG. 2, the location relevant to the vibration peak at the time of inner ring damage is a broken line, and the location relevant to the vibration peak at the time of outer ring damage is a solid line, and the annotation is described. In FIG. 3, the points related to the vibration peak at the time of rolling element damage are noted with a solid line.

When the inner ring is damaged, as shown in FIG. 2, two vibration peaks (peak pairs) separated by twice the same specific vibration source rotational frequency (= 1) exist at a plurality of places. Furthermore, since the specific frequency of the average of each peak pair is always a natural number multiple of the ratio fundamental frequency, even if the ratio fundamental frequency is unknown, the ratio frequency of the average of the peak pair is a natural number ratio It can be convinced that it is the vibration peak by inner ring damage by confirming.
When the outer ring is damaged, as shown in FIG. 2, vibration peaks are present at a plurality of places such that the ratio of the specific frequency becomes a natural number ratio. It is generally designed that the relative fundamental frequency of bearings is not a natural number, and furthermore, if some of the relative frequencies of their vibration peaks are not natural numbers, it can be inferred that the relative fundamental frequency is not a natural number. It can be convinced that it is the vibration peak due to the outer ring damage.
When the rolling element is damaged, as shown in FIG. 3, two vibration peaks (peak pairs) separated by twice the same specific vibration source rotational frequency exist at a plurality of places. Furthermore, since the specific frequency of the average of each peak pair is always a natural number multiple of the ratio fundamental frequency, even if the ratio fundamental frequency is unknown, the ratio frequency of the average of the peak pair is a natural number ratio It can be convinced that it is the vibration peak by rolling element damage by confirming. In addition, since the value of the specific vibration source rotational frequency is always smaller than 1, discrimination from inner ring damage is possible.

FIG. 4 shows a flowchart of a method of diagnosing a rolling bearing abnormality, which will be specifically described based on the flowchart.
First, one of all the conditions of the diagnostic rotational frequency registered in advance is displayed on the display / operation unit 13 as shown in FIG. 7 (S1).
Next, the user of the diagnostic device commands the displayed diagnostic rotational frequency to the control device 6 (S2).
Next, the user of the diagnostic apparatus checks the display unit 8 to determine whether the rotating body supported by the bearing to be diagnosed is rotating at the rotational frequency displayed on the display / operation unit 13 If it does, the start of measurement is instructed from the display / operation unit 13 (S3).
Then, the abnormality diagnosis apparatus 200 measures and records the vibration acceleration (S4), determines that the measurement is completed when the required data length is reached, and shifts to S6 (S5). In S6, the rotational frequency and vibration acceleration at the time of vibration measurement are correlated and recorded.
In S7, it is determined whether measurement of all conditions of the diagnostic rotational frequency registered in advance is completed, and if measurement of all conditions is completed, the process proceeds to S8. If not completed, the process returns to S1 to display another diagnostic rotational frequency whose measurement has not been completed. Steps S4 to S7 here are vibration measurement steps.

In S8, first, the vibration acceleration is subjected to Fourier transform to calculate an amplitude spectral density. Next, a corrected vibration value is calculated by dividing the absolute value of the amplitude spectral density by the square of the angular frequency at measurement (the rotational frequency at measurement multiplied by 2 and the ratio of the circle). The rotation synchronization component emphasized waveform is calculated by calculating the average of the frequency components having the same relative frequency of the corrected vibration value at a plurality of rotation frequencies (rotation synchronization component emphasis waveform calculation step). If the relative frequency resolutions do not match, interpolation processing is performed to unify the relative frequency resolutions to calculate an average.
Next, in S9, assuming the ratio fundamental frequency and the specific vibration source rotational frequency, it is determined whether or not there is a vibration peak at the position of ratio basic frequency × natural number ± specific vibration source rotational frequency of the rotation synchronization component emphasis waveform. The judgment as to whether or not a vibration peak exists in a specific frequency is made by comparing the moving average of the rotation synchronization component emphasis waveform and the rotation synchronization component emphasis waveform, and if the value of the rotation synchronization component emphasis waveform is larger, the relative frequency It may be determined that a vibration peak exists in Then, a ratio (a coincidence ratio) in which a vibration peak exists at the position of the ratio fundamental frequency × natural number ± specific vibration source rotational frequency is calculated.

Next, in S10, it is determined whether the search of all combinations of the ratio fundamental frequency of the predetermined search range and the specific vibration source rotational frequency is completed, and if completed, the process proceeds to S11.
In S11, types of bearing abnormalities (inner ring damage, rolling element damage, outer ring damage) according to the value of the specific vibration source rotational frequency for the ratio fundamental frequency and the specific vibration source rotational frequency whose coincidence rate is equal to or higher than the predetermined threshold To judge. S9 to S11 here are bearing damage type determination steps.
Then, in S12, the diagnostic result as shown in FIG. Here, since the coincidence rate is calculated to be high when the specific vibration source rotational frequency is 0 corresponding to the outer ring and 1 corresponding to the inner ring, it is displayed that the outer ring and the inner ring may be damaged.

FIG. 5 shows a flowchart of another method for diagnosing a rolling bearing abnormality, which will be specifically described based on this flowchart.
Steps S1 to S8 are the same as the method of FIG. First, one of all the conditions of the diagnostic rotational frequency registered in advance is displayed on the display / operation unit 13 as shown in FIG. 7 (S1). The user of the diagnostic device commands the displayed diagnostic rotational frequency to the control device 6 (S2). The user of the diagnostic apparatus checks the display unit 8 to determine whether the rotating body supported by the bearing to be diagnosed is rotating at the rotational frequency displayed on the display / operation unit 13 and agrees. In the case, the start of measurement is instructed from the display / operation unit 13 (S3). The vibration acceleration is measured and recorded (S4), and when the required data length is reached, it is determined that the measurement is completed, and the process proceeds to S6 (S5). The rotational frequency at the time of vibration measurement and vibration acceleration are correlated and recorded (S6). It is determined whether measurement of all conditions of the diagnostic rotational frequency registered in advance is completed, and if measurement of all conditions is completed, the process proceeds to S8. If not completed, the process returns to S1 to display another diagnostic rotational frequency whose measurement has not been completed.
In S8, first, the vibration acceleration is subjected to Fourier transform to calculate an amplitude spectral density. The corrected vibration value is calculated by dividing the absolute value of the amplitude spectral density by the square of the angular frequency at the time of measurement (the value obtained by multiplying the rotational frequency at the time of measurement by 2 and the circle ratio). The rotation synchronization component emphasized waveform is calculated by calculating the average of the frequency components with the same relative frequency of the corrected vibration value at a plurality of rotation frequencies. If the relative frequency resolutions do not match, interpolation processing is performed to unify the relative frequency resolutions to calculate an average.

Then, in S109, first, four relative frequencies determined to have vibration peaks are selected. The judgment as to whether or not a vibration peak exists at a specific frequency is made by comparing the rotation synchronization component emphasis waveform with the moving average of the rotation synchronization component emphasis waveform, and if the value of the rotation synchronization component emphasis waveform is larger, the ratio It may be determined that a vibration peak exists in the frequency.
Next, in S110, combinations of the specific frequencies F ₁ , F ₂ , F ₃ , and F _{4 of} the four selected vibration peaks and different natural numbers N ₁ and N ₂ having no common divisor are the following It is determined whether the relational expression of Equation 2 is satisfied. _{Incidentally,} the vibration of the _{F 1, F 2, F 3} , the outer ring damaged when may be chosen the same ratio frequency _{(F 1} and _F 2, _{F 3} and _{F 4} are identical ratio frequency Each of the _{F 4} Peak can be detected).

In the discrimination of S110, the combination of the specific frequencies F ₁ , F ₂ , F ₃ and F _{4 of four} vibration peaks and different natural numbers N ₁ and N ₂ having no common divisor satisfies the relational expressions of the numbers 1 and 2 In the case, the process proceeds to step S111, and if the relational expression is not satisfied, the process returns to step S109, four different specific frequencies are selected, and the determination in step S110 is performed again.
In S111, first, the ratio fundamental frequency R ₀ and the relative vibration source rotational frequency R ₁ are calculated from the following Equation 3 and Equation 4 for the combination of the specific frequencies of the vibration peaks that satisfy the relational expressions of Equations 1 and 2.

Then, the type of bearing abnormality (inner ring damage, rolling element damage, outer ring damage) is determined according to the calculated value of the specific vibration source rotational frequency R ₁ , and the diagnosis result as shown in FIG. Display on section 13. Here, since the specific vibration source rotational frequency is calculated to be 0.45 corresponding to the rolling element, it is displayed that the rolling element may be damaged. S109 to S111 here are bearing damage type determination steps.

FIG. 6 shows a flowchart of another method for diagnosing abnormality of a rolling bearing using machine learning, which will be specifically described based on this flowchart.
Here, a rotation synchronization component emphasis waveform calculated by a method described later for a bearing whose bearing abnormality is known is input, and inner ring damage (1) or not (0), rolling element damage (1) or not (0), whether there is outer ring damage (1) or not (0), machine learning is performed in advance in an external learning device (not shown in FIG. 1) using teacher data as an output. The model is provided in the calculation unit 12 of FIG. For example, a multi-layered neural network can be used as the form of the mathematical model. Based on the distribution of output when the value of evaluation data (rotation synchronization component emphasis waveform) is input for a bearing whose bearing abnormality is known, the threshold value used to determine whether the inner ring, rolling elements, and outer ring are abnormal is determined in advance. It is provided in the calculation unit 12.

Steps S1 to S8 are the same as the method of FIG. One of all conditions of the diagnostic rotational frequency registered in advance is displayed on the display / operation unit 13 as shown in FIG. 7 (S1). The user of the diagnostic device commands the displayed diagnostic rotational frequency to the control device 6 (S2). The user of the diagnostic apparatus checks the display unit 8 to determine whether the rotating body supported by the bearing to be diagnosed is rotating at the rotational frequency displayed on the display / operation unit 13 and agrees. In the case, the start of measurement is instructed from the display / operation unit 13 (S3). The vibration acceleration is measured and recorded (S4), and when the required data length is reached, it is determined that the measurement is completed, and the process proceeds to S6 (S5). The rotational frequency at the time of vibration measurement and vibration acceleration are correlated and recorded (S6). It is determined whether measurement of all conditions of the diagnostic rotational frequency registered in advance is completed, and if measurement of all conditions is completed, the process proceeds to S8. If not completed, the process returns to S1 to display another diagnostic rotational frequency whose measurement has not been completed.
In S8, first, the vibration acceleration is subjected to Fourier transform to calculate an amplitude spectral density. The corrected vibration value is calculated by dividing the absolute value of the amplitude spectral density by the square of the angular frequency at the time of measurement (the value obtained by multiplying the rotational frequency at the time of measurement by 2 and the circle ratio). The rotation synchronization component emphasized waveform is calculated by calculating the average of the frequency components with the same relative frequency of the corrected vibration value at a plurality of rotation frequencies. The moving averages of the rotation synchronization component emphasized waveform and the rotation synchronization component emphasized waveform are compared, and if the value of the rotation synchronization component emphasized waveform is larger, it is determined that a vibration peak is present in the relative frequency. If the relative frequency resolutions do not match, interpolation processing is performed to unify the relative frequency resolutions to calculate an average.

Then, in S209, the rotation synchronization component emphasis waveform is input to the learned mathematical model, and the output values of whether or not the inner ring is damaged, whether the rolling element is damaged, and whether the outer ring is damaged are calculated. Yes (bearing damage type identification step).
If the value of each output value exceeds the preset threshold value in S209, it is judged as abnormal, and the diagnostic result as shown in FIG. 10 is displayed on the display / operation unit 13 in S210. In FIG. 10, the output value from the learned mathematical model, the threshold value for judging whether it is abnormal, and the judgment result whether it is abnormal or not are judged as graphs as the rotation synchronization component emphasis waveform input to the learned mathematical model. Is displayed.

As described above, according to the abnormality diagnosis method and the abnormality diagnosis apparatus 200 of the above-described embodiments and the abnormality diagnosis program, a rotation synchronization component emphasis waveform emphasizing a vibration component of a frequency proportional to the rotation frequency of the spindle 1 with respect to the measured vibration. Since the type of damage in the bearing 7 is determined based on the regularity of the vibration peak position of the rotation synchronization component emphasis waveform, the type of bearing abnormality is determined from the regularity of the vibration peak without using bearing specifications. It becomes possible.
Further, in the present invention, since the ratio fundamental frequency and the specific vibration source rotational frequency are estimated, the fundamental frequency and the characteristic frequency can be calculated. For this reason, it is also possible to diagnose the conventional diagnostic technology in which it is necessary to use bearing specifications to calculate the fundamental frequency and the characteristic frequency. If the fundamental frequency at a certain rotational frequency is required, it can be calculated by multiplying the value of the ratio fundamental frequency by the rotational frequency. A specific vibration source of a combination of a ratio fundamental frequency determined as rolling element damage and a specific vibration source rotational frequency when a frequency at which a rolling element cage rotates (a rolling element revolves) at a certain rotation frequency is required It is possible to calculate by multiplying the value of the rotational frequency by the rotational frequency.

  Particularly in this case, in the vibration measurement step, the vibration is measured at a plurality of different rotational frequencies, and in the rotation synchronization component emphasized waveform calculation step, the vibration is frequency analyzed to calculate the magnitude of the vibration for each frequency. Since the magnification of vibration frequency (specific frequency) is equal to the average of vibration magnitude for the same component (S8), the influence of disturbance is further reduced and it becomes easy to capture the regularity of the vibration peak. The estimation accuracy of the bearing abnormality is improved.

Further, in the abnormality diagnosis method shown in FIG. 6, a machine learned using teacher data having as input the rotation synchronization component emphasis waveform, and outputting at least one of the presence or absence of the inner ring flaw, the presence or absence of the rolling body flaw, and the presence or absence of the outer ring flaw. Since the type of bearing damage is determined by inputting the rotation synchronization component emphasis waveform using a learning model, it is possible to determine the type of bearing abnormality from the regularity of vibration peak more generally than the process considered by humans. .
Then, in the abnormality diagnosis apparatus 200, since the ratio fundamental frequency corresponding to the bearing damage and the specific vibration source rotational frequency are displayed on the display / operation unit 13, the other bearing is used using the value of the frequency displayed as a numeral. It is possible to perform diagnostic processing.
Furthermore, here, since the position of the specific ratio frequency and the rotation synchronization component emphasizing waveform are combined and displayed on the display / operation unit 13, it is possible to visually grasp whether or not the regularity of the vibration peak can be correctly extracted. The validity of the diagnostic results can be easily verified. In addition, the specific frequency to be focused on in each type of damage can be easily understood.

In each of the above examples, an example is shown in which the rotational frequency of the rotating body supported by the bearing to be diagnosed can be directly instructed. However, when diagnosing a support bearing of a motor that rotates a machine tool spindle via a belt The rotational frequency of the support bearing of the motor may be designated to be the diagnostic frequency in consideration of the reduction ratio.
In addition, by matching the rotational frequency of the bearing to be diagnosed with the diagnostic rotational frequency displayed on the display / operation unit 13, the method of storing the vibration acceleration and the rotational frequency in association with each other has been shown. The frequency may be input by the user of the diagnostic device from the display / operation unit 13 or may be a method by which the abnormality diagnostic device 200 can directly capture the information of the speed detector 5.
Furthermore, the rotational frequency (often the unit is Hz) and the rotational speed (in the machine tool spindle, the unit min ^-1 is often used) are the same quantities in the relationship of ¹ Hz = 60 min ^-1 . You may use either.

On the other hand, using the rolling element diameter d, the pitch circle diameter D, and the contact angle α, the ratio of the rolling element revolution frequency to the rotational frequency in the absence of slip (specific vibration source rotational frequency R _{1 in} the case of rolling element abnormality) It is expressed by the following equation 5.

From equation 5, it can be seen that the specific vibration source rotational frequency in the case of rolling element abnormality decreases as the ratio of the rolling element diameter to the pitch circle diameter increases. Since at least three rolling elements are required to form a rolling bearing, the ratio of the rolling element diameter to the pitch circle diameter is maximized in a bearing having a structure in which three rolling elements are closely arranged. The lower limit value of the specific vibration source rotational frequency is (1−√3 ÷ 2) ÷ 2 ÷ 0.0669. From equation 5, it can be seen that when the rolling element diameter is made infinitesimal, the specific vibration source rotational frequency in the case of rolling element abnormality becomes maximum. Therefore, geometrically, the upper limit value of the relative vibration source rotational frequency is 0.5. The range which can be taken by the specific vibration source rotational frequency in the case of the rolling element abnormality is (1− ÷ 3 ÷ 2) / 2 or more and 0.5 or less at the widest estimate.
In the case of a bearing used for a machine tool spindle, the specific vibration source rotational frequency in the case of rolling element abnormality is around 0.45, so the assumed specific vibration source rotational frequency is 0, 0.4 or more and 0.5 or less, 1 The search may be further limited, for example.

Also, when calculating a rotation synchronization component emphasis waveform from vibration acceleration measured at a plurality of rotation frequencies, the square frequency of the angular frequency at measurement (the value obtained by multiplying the rotation frequency at measurement by 2 and the circle ratio) Although an example of dividing processing is shown, it is possible to emphasize the vibration component of the frequency proportional to the rotation frequency of the rotating body and omit it if the purpose of enabling easy extraction of the vibration peak can be achieved. You may add another process.
Furthermore, the same argument holds true even if all of the relative frequency, relative fundamental frequency, relative vibration source rotational frequency, and values to be compared with these values are multiplied by the same value. That is, for example, determining whether or not the specific vibration source rotational frequency matches 1 is equivalent to determining whether or not the specific vibration source rotational frequency × 30 Hz matches 1 × 30 Hz.

Besides, in the present embodiment, the machine tool and the abnormality diagnosis device are shown separately, but the abnormality diagnosis device may be incorporated in the control device.
Also, the abnormality diagnosis device can communicate with multiple machine tools by wire or wireless, measure vibration on the machine tool side and acquire data, and execute the abnormality diagnosis method for each machine tool based on the abnormality diagnosis program. You may do it. Furthermore, the measurement data of the vibration may be stored in the storage unit and the abnormality diagnosis may be performed by the abnormality diagnosis program at a predetermined timing without being limited to the case of continuously performing the measurement of the vibration to the abnormality diagnosis. .

  1 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · speed control device · · · · control unit · · · · · · · · · display unit, 9 · · · vibration sensor 10... A / D converter, 11 .. storage unit, 12 .. arithmetic unit, 13 .. display and operation unit, 100 .. machine tool, 200 .. abnormality diagnosis device.

Claims (8)

A method of diagnosing an abnormality in a rolling bearing supporting a rotating body, the method comprising:
A vibration measurement step of measuring the vibration of the rotating body a plurality of times ;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency A rotation synchronization component emphasis waveform calculation step of calculating an average of the ratio frequencies for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
Based on the regularity of the vibration peak position of the rotation synchronization component emphasis waveform , a specific vibration source rotation frequency which is a ratio of a vibration source rotation frequency which is a frequency at which a force generation direction changes in the rolling bearing If the value of the specific vibration source rotational frequency is 1, it is determined that the inner ring of the rolling bearing is damaged, and the value of the specific vibration source rotational frequency is the rotational frequency of the rolling elements of the rolling bearing. In the case of the possible range of the value divided by the frequency, it is judged that the rolling element is damaged, and when the value of the specific vibration source rotational frequency is 0, the bearing damage judged as the outer ring damage of the rolling bearing Type determination step;
A method for diagnosing abnormality of a rolling bearing, characterized in that: The matching rate is the ratio of the presence of a vibration peak at a specific ratio frequency calculated assuming the ratio fundamental frequency and the ratio vibration source rotation frequency in the rotation synchronization component emphasis waveform in the estimation of the ratio vibration source rotation frequency. Is calculated, and if the coincidence rate exceeds a predetermined threshold value, the rotation synchronization component emphasis waveform has regularity represented by the assumed ratio fundamental frequency and the assumed relative vibration source rotational frequency The method for diagnosing abnormality of a rolling bearing according to claim 1 , wherein it is determined that the specific vibration source rotational frequency is assumed. Wherein the vibration measuring step, the abnormality diagnostic method of a rolling bearing according to claim 1 or 2, characterized in that to measure the vibration at different rotational frequencies. A method of diagnosing an abnormality in a rolling bearing supporting a rotating body, the method comprising:
A vibration measurement step of measuring the vibration of the rotating body a plurality of times;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency A rotation synchronization component emphasis waveform calculation step of calculating an average of the ratio frequencies for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
The rotation synchronization component enhancement is performed using a machine learning model learned using teacher data in which the rotation synchronization component enhancement waveform is input and at least one of inner ring injury, rolling body injury, and outer ring injury is output. A bearing damage type determination step of inputting a waveform to determine the type of damage to the rolling bearing;
A method for diagnosing abnormality of a rolling bearing, characterized in that: An apparatus for diagnosing an abnormality in a rolling bearing supporting a rotating body, the apparatus comprising:
Vibration measuring means for measuring the vibration of the rotating body a plurality of times ;
The vibration is subjected to frequency analysis to obtain the magnitude of the vibration for each frequency, and the dimension of the vibration obtained by each measurement using the dimensionless amount obtained by dividing the frequency of the vibration by the rotational frequency as a relative frequency Rotation synchronization component emphasis waveform calculation means for calculating an average of the relative frequency for each of the ratio frequencies as a rotation synchronization component emphasis waveform;
Based on the regularity of the vibration peak position of the rotation synchronization component emphasis waveform , a specific vibration source rotation frequency which is a ratio of a vibration source rotation frequency which is a frequency at which a force generation direction changes in the rolling bearing If the value of the specific vibration source rotational frequency is 1, it is determined that the inner ring of the rolling bearing is damaged, and the value of the specific vibration source rotational frequency is the rotational frequency of the rolling elements of the rolling bearing. In the case of the possible range of the value divided by the frequency, it is judged that the rolling element is damaged, and when the value of the specific vibration source rotational frequency is 0, the bearing damage judged as the outer ring damage of the rolling bearing Type determination means;
An abnormality diagnosis device for a rolling bearing comprising: Said bearing damage type determination means, the ratio fundamental frequency corresponding to the damage of the rolling bearing, and the ratio vibration source rotational frequency, and the specific ratios frequencies calculated from said ratio vibration source rotation frequency as the ratio fundamental frequency While confirming the existence of the vibration peak position for each type of damage of the rolling bearing, and determining the type of damage based on the value of the specific vibration source rotational frequency ,
6. The abnormality diagnosis device for a rolling bearing according to claim 5 , further comprising display means for displaying the relative fundamental frequency and the relative vibration source rotational frequency together. Said bearing damage type determination means, the ratio fundamental frequency corresponding to the damage of the rolling bearing, and the ratio vibration source rotational frequency, and the specific ratios frequencies calculated from said ratio vibration source rotation frequency as the ratio fundamental frequency The presence of the vibration peak value for each type of damage to the rolling bearing is confirmed using the above , and the type of damage is determined based on the value of the specific vibration source rotational frequency ,
6. The apparatus according to claim 5 , further comprising display means for displaying the position of the specific ratio frequency and the rotation synchronization component emphasizing waveform together. The rotation synchronous component emphasis waveform calculation step and bearing according to any one of claims 1 to 4 , wherein the vibration of the rotating body measured at a predetermined rotational frequency is input together with the rotational frequency. An abnormality diagnosis program for a rolling bearing, characterized in that the damage type determination step is executed.

Images