Damage condition diagnosing system and damage condition diagnosing method

Abstract

To provide a technique that can diagnose the damage condition which indicates the degree of progression of damage of a rolling bearing with high accuracy.SOLUTION: A sensor (20) detects vibration generated when a rolling bearing (10) is rotated. A diagnosing device (100) includes a preprocessing unit (30), a feature extracting unit (50), and a damage condition diagnosing unit (60). The preprocessing unit (30) generates a vibration waveform by acquiring an output signal of the sensor (20) over a period during which a rotating race (12) rotates a plurality of times. The feature extracting unit (50) extracts a plurality of features from the vibration waveform. The damage condition diagnosing unit (60) diagnoses the damage condition of the rolling bearing (10) on the basis of the plurality of features extracted by the feature extracting unit. The plurality of features include: a peeling length or a non-peeling length in the circumferential direction of the rolling bearing (10); a peeling pass count, which is the number of times each rolling element (13) passes through peeling during the time taken for the rotating race (12) to rotate once; and a variability index indicating variability in the amplitude of the vibration waveform.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a damage condition diagnosis system and a damage condition diagnosis method for diagnosing the damage condition of a rolling bearing. [Background technology]

[0002] Vibration characteristics are used to diagnose the damage state, which indicates the extent of damage to rolling bearings used in machinery and equipment. For example, Japanese Patent Application Laid-Open No. 2017-32520 (Patent Document 1) discloses a technology that creates waveforms in three frequency bands from the vibration data of a rolling bearing and diagnoses the damage state of the rolling bearing from the vibration effective value for each frequency band. In the early stages of damage, only the effective value of the higher frequency band of the three frequency bands reacts, and as the damage progresses, the effective value of the lower frequency band increases. In Patent Document 1, the damage state of the rolling bearing is evaluated on a three-level scale by monitoring the vibration effective values ​​of these three frequency bands.

[0003] Furthermore, Japanese Patent No. 6733838 (Patent Document 2) discloses a technology for estimating the length of a flake in the circumferential direction of a rolling bearing, based on the time difference between the time when a rolling element enters the flake in the raceway ring and the time when the rolling element exits the flake in the raceway ring, in regard to damage caused by the flake. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2017-32520 [Patent Document 2] Patent No. 6733838 Summary of the Invention [Problem to be solved by the invention]

[0005] The technique described in Patent Document 1 evaluates the damage state of a rolling bearing in three stages, which has the problem that it is not possible to quantitatively evaluate the damage state.

[0006] On the other hand, the technology described in Patent Document 2 makes it possible to quantitatively evaluate the state of damage from the length of spalling in the circumferential direction of the rolling bearing. However, Patent Document 2 is limited to evaluating the progression of spalling that has occurred in one location on the rolling bearing, and is not able to evaluate spalling that has occurred in multiple locations on the rolling bearing. Furthermore, Patent Document 2 does not make it possible to estimate where spalling has occurred on the rotating ring, fixed ring, or rolling elements that make up the rolling bearing.

[0007] The present disclosure has been made to solve the above-mentioned problems, and an object of the present disclosure is to provide a technology that can diagnose with high accuracy the damage state that indicates the degree of progression of damage to a rolling bearing from the occurrence of one flake to the occurrence of multiple flakes. [Means for solving the problem]

[0008] A damage state diagnosis system according to one aspect of the present disclosure diagnoses a damage state indicating the extent of damage progression in a rolling bearing. The rolling bearing includes a fixed ring, a rotating ring, and a plurality of rolling elements circumferentially arranged at equal pitch intervals between the fixed ring and the rotating ring so as to be able to roll. The damage state diagnosis system includes a sensor that detects vibrations generated when the rolling bearing is rotated, and a diagnosis device that diagnoses the damage state of the rolling bearing based on an output signal from the sensor. The diagnosis device includes a preprocessing unit, a feature extraction unit, and a damage state diagnosis unit. The preprocessing unit generates a vibration waveform by acquiring the output signal from the sensor over a period of time during which the rotating ring rotates multiple times. The feature extraction unit extracts a plurality of feature amounts from the vibration waveform. The damage state diagnosis unit diagnoses the damage state of the rolling bearing based on the plurality of feature amounts extracted by the feature extraction unit. The multiple feature quantities include the length of the flake or non-flake in the circumferential direction of the rolling bearing, the number of flake passes each rolling element passes through the flake within the time it takes for the rotating wheel to make one rotation, and a variation index representing the variation in the amplitude of the vibration waveform. [Effects of the Invention]

[0009] According to the present disclosure, it is possible to diagnose with high accuracy the damage state that indicates the degree of progression of damage to a rolling bearing from the occurrence of one flaking to the occurrence of multiple flaking. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram showing an overall configuration of a damage state diagnosis system according to an embodiment of the present disclosure. [Figure 2] FIG. 2 illustrates an example of a hardware configuration of a diagnostic device. [Figure 3] FIG. 10 is a diagram showing an example of a displacement vibration waveform. [Figure 4] FIG. 10 is a diagram showing an example of a displacement vibration waveform during one rotation of a rotating wheel. [Figure 5] FIG. 10 is a diagram showing an example of a displacement vibration waveform during one rotation of a rotating wheel. [Figure 6] FIG. 2 is a diagram schematically illustrating a load zone of a rolling bearing. [Figure 7] FIG. 10 is a schematic diagram for explaining pitch separation. [Figure 8] FIG. 10 is a diagram showing an example of a displacement vibration waveform during four rotations of a rotating wheel. [Figure 9] FIG. 10 is a diagram showing an example of a displacement vibration waveform during four rotations of a rotating wheel. [Figure 10] FIG. 10 is a diagram showing an example of a displacement vibration waveform for two rotations when pitch interval separation occurs in the fixed ring. [Figure 11] FIG. 10 is a diagram showing an example of a displacement vibration waveform for two rotations when pitch interval peeling occurs on the rotating wheel. [Figure 12] 10 is a flowchart illustrating a flow of a damage state diagnosis process in a damage state diagnosis unit. [Figure 13] FIG. 2 is a diagram showing the configuration and operating conditions of the rolling bearing used in Example 1. [Figure 14] FIG. 1 is a diagram showing peeling conditions in Example 1. [Figure 15] FIG. 10 is a diagram showing the number of peel passes extracted from the displacement vibration waveform under each condition. [Figure 16] FIG. 10 is a diagram showing variation indexes extracted from displacement vibration waveforms under each condition. [Figure 17] FIG. 10 is a diagram showing the peeling length extracted from the displacement vibration waveform under each condition. [Figure 18] FIG. 10 shows the change in the ratio of static displacement to the number of peels. [Figure 19] FIG. 10 is a diagram showing the results of calculating the variation index when n=4. [Figure 20] FIG. 10 is a diagram showing an example of a displacement vibration waveform during four rotations of a rotating wheel. [Figure 21] FIG. 10 is a diagram showing an example of a displacement vibration waveform during four rotations of a rotating wheel. [Figure 22] FIG. 10 shows a modified variability index. [Figure 23] FIG. 10 is a block diagram showing a modified example of the damage state diagnosis system according to the present embodiment. [Figure 24] FIG. 10 is a diagram showing an example of a displacement vibration waveform. [Figure 25] 10 is a flowchart illustrating a flow of a damage state diagnosis process in a damage state diagnosis unit according to a modified example. [Figure 26] FIG. 10 is a diagram showing the peeling conditions in Example 2. [Figure 27] FIG. 10 is a diagram showing the number of peel passes extracted from the displacement vibration waveform under each condition. [Figure 28] FIG. 10 is a diagram showing variation indexes extracted from displacement vibration waveforms under each condition. [Figure 29] FIG. 10 is a diagram showing the non-peeling length extracted from the displacement vibration waveform under each condition. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In the drawings, the same or corresponding parts are designated by the same reference numerals, and description thereof will not be repeated.

[0012] <Configuration of damage condition diagnosis system> Fig. 1 is a block diagram showing the overall configuration of a damage state diagnosis system according to an embodiment of the present disclosure. As shown in Fig. 1, the damage state diagnosis system according to the present embodiment is a system for diagnosing a damage state that indicates the extent of progress of damage to a rolling bearing 10.

[0013] The rolling bearing 10 includes a fixed ring 11, a rotating ring 12, and a plurality of rolling elements 13 arranged to be able to roll between the fixed ring 11 and the rotating ring 12. The rolling bearing 10 further includes a cage 14 that holds the plurality of rolling elements 13 so that they can roll freely. The plurality of rolling elements 13 are arranged at equal pitch intervals in the circumferential direction of the rolling bearing 10 by the cage 14.

[0014] Rotating ring 12 is fitted onto the outside of a rotating shaft of mechanical equipment. Fixed ring 11 is fitted inside a housing or the like. Note that in this embodiment, a case where the rolling bearing is constituted by a roller bearing whose inner ring is a rotating ring will be described as a representative example, but the scope of application of this disclosure is not limited to this type of bearing damage estimation device, and the rolling bearing may be one whose outer ring is a rotating ring, or may be a ball bearing, etc.

[0015] The damage state diagnosis system includes a sensor 20 and a diagnosis device 100. The sensor 20 is configured to detect the acceleration, velocity, displacement, or the like of vibrations that occur when the rolling bearing 10 is rotated. An acceleration sensor, velocity sensor, or displacement sensor can be used as the sensor 20. The sensor 20 outputs a signal indicating the detected value to the diagnosis device 100.

[0016] The diagnostic device 100 diagnoses the state of damage of the rolling bearing 10 based on the output signal of the sensor 20. Fig. 2 is a diagram showing an example of the hardware configuration of the diagnostic device 100. As shown in Fig. 2, the diagnostic device 100 is configured to include a CPU (Central Processing Unit) 101, a RAM (Random Access Memory) 102, a ROM (Read Only Memory) 103, an I / F (Interface) device 104, and a storage device 105. The CPU 101, RAM 102, ROM 103, I / F device 104, and storage device 105 exchange various types of data via a communication bus 106.

[0017] The CPU 101 loads a program stored in the ROM 103 into the RAM 102 and executes it. The program stored in the ROM 103 describes the processes to be executed by the diagnostic device 100.

[0018] The I / F device 104 is an input / output device for exchanging signals and data with the outside of the diagnostic device 100 (for example, the sensor 20, a monitoring terminal, etc.). The I / F device 104 receives an output signal from the sensor 20. The I / F device 104 also transmits damage state data indicating the diagnosis result to the monitoring terminal.

[0019] The storage device 105 is a storage for storing various types of information, and stores information on the rolling bearing 10 to be diagnosed, information on the vibration waveform generated from the output signal of the sensor 20, information on features extracted from the vibration waveform, etc. The storage device 105 is, for example, a hard disk drive (HDD) or a solid state drive (SSD).

[0020] 2, at least a part of the diagnostic device 100 can be configured using circuits such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit).Furthermore, at least a part of the diagnostic device 100 can also be configured using analog circuits.

[0021] Returning to Fig. 1, the diagnostic device 100 includes a preprocessing unit 30, a storage unit 40, a feature extraction unit 50, and a damage state diagnosis unit 60. The functions of each unit can be realized by at least one of software processing by the CPU 101 and hardware processing. The functions and operations of each unit of the diagnostic device 100 will be described below.

[0022] (Pre-processing unit 30) The pre-processing unit 30 generates a vibration waveform by acquiring the output signal of the sensor 20 over a period of time during which the rotating wheel 12 rotates multiple times. In this embodiment, the pre-processing unit 30 generates a displacement vibration waveform represented by a displacement based on the displacement signal detected by the sensor 20. The pre-processing unit 30 may generate a velocity vibration waveform represented by a velocity based on the velocity signal detected by the sensor 20, or may generate an acceleration vibration waveform represented by an acceleration based on the acceleration signal detected by the sensor 20.

[0023] (Storage unit 40) The storage unit 40 stores the vibration waveform (for example, a displacement vibration waveform) generated by the preprocessing unit 30.

[0024] (Feature extraction unit 50) The feature extraction unit 50 extracts a plurality of feature values ​​from the vibration waveform (e.g., displacement vibration waveform) stored in the storage unit 40. In this specification, a "feature value" is a quantitative numerical representation of a feature of the vibration waveform (e.g., displacement vibration waveform).

[0025] The multiple feature quantities include a flaking length D in the circumferential direction of the rolling bearing 10 (hereinafter also referred to as "flaking length D"), the number of times N that the rolling element 13 passes through the flaking within the time it takes for the rotating ring 12 to make one rotation (i.e., the rotation period of the rotating ring 12) (hereinafter also referred to as "flaking pass number N"), and an index V that represents the variation in the amplitude of the vibration waveform (hereinafter also referred to as "variation index V"). Below, the definition of each feature quantity and the method of extracting it will be explained.

[0026] (1) Peel length D Fig. 3 is a diagram showing an example of a displacement vibration waveform. Fig. 3 shows a displacement vibration waveform when one flaking occurs on the raceway surface of fixed ring 11. The vertical axis of Fig. 3 represents vibration displacement, and the horizontal axis represents time.

[0027] As shown in Figure 3, when the rolling element 13 enters a flake on the fixed ring 11, the displacement vibration waveform has a positive peak. When the rolling element 13 escapes from the flake, the displacement vibration waveform has a negative peak. In contrast to Figure 3, there are also cases where the displacement vibration waveform has a negative peak when the rolling element 13 enters a flake and a positive peak when the rolling element 13 escapes from the flake. In either case, a pair of positive and negative peaks associated with the entry and exit of the rolling element 13 into and from the flake appear in the displacement vibration waveform at the pitch intervals of the rolling elements 13. The amplitude threshold for detecting entry into and exit from a flake can be determined, for example, from the quartile of the amplitude of the vibration waveform or a multiple of the root mean square (rms) value of the normal value.

[0028] If the operating speed of the rolling bearing 10 is constant, the flake length D can be estimated based on the time difference between the time when the rolling element 13 enters the flake and the time when the rolling element 13 exits the flake. Then, from the estimated flake length D, the number of rolling elements 13 passing through the flake (hereinafter also referred to as the "number of passing rolling elements") can be estimated.

[0029] Specifically, as shown in Figure 3, when the peeling length D is equal to or less than the pitch interval of the rolling elements 13, the number of passing rolling elements will be 0 to 1. Although not shown, when the peeling length D exceeds the pitch interval of the rolling elements 13, the number of passing rolling elements will be 1 to 2, and there may always be rolling elements 13 that do not bear any load. In this way, by comparing the peeling length D with the pitch interval of the rolling elements 13, the number of passing rolling elements can be estimated.

[0030] (2) Number of peel passes N 4 and 5 are diagrams showing examples of displacement vibration waveforms in the time it takes for the rotating ring 12 to make one rotation. Fig. 4 shows a displacement vibration waveform when one peeling has occurred on the raceway surface of the fixed ring 11. Fig. 5 shows a displacement vibration waveform when one peeling has occurred on the raceway surface of the rotating ring 12. The vertical axis of Figs. 4 and 5 shows vibration displacement, and the horizontal axis shows the number of rotations of the rotating ring 12.

[0031] 4, when one flake occurs on the fixed ring 11, a pair of positive and negative peaks associated with the entry and exit of the rolling element 13 into and out of the flake appear in the displacement vibration waveform at the pitch intervals of the rolling element 13. It can be seen that the multiple positive peaks appearing at the pitch intervals of the rolling element 13 have a substantially constant magnitude (amplitude), and the multiple negative peaks appearing at the pitch intervals of the rolling element 13 also have a substantially constant magnitude (amplitude).

[0032] As shown in Figure 5, even when one flake has occurred on the rotating ring 12, a pair of positive and negative peaks associated with the rolling element 13's entry into and exit from the flake appear in the displacement vibration waveform at the pitch intervals of the rolling element 13. However, unlike Figure 4, it can be seen that the pair of positive and negative peaks associated with the rolling element 13's entry into and exit from the flake occur only when the flake on the rotating ring 12 enters the load zone. Also, unlike Figure 4, it can be seen that the multiple positive peaks that appear at the pitch intervals of the rolling element 13 are amplitude modulated, and the multiple negative peaks that appear at the pitch intervals of the rolling element 13 are amplitude modulated.

[0033] Fig. 6 is a diagram showing a schematic diagram of the load zone of rolling bearing 10. In Fig. 6, rotating ring 12 rotates together with the rotating shaft of the mechanical equipment, and multiple rolling elements 13 roll between rotating ring 12 and fixed ring 11. At this time, each rolling element 13 revolves while rotating on its own axis.

[0034] In this state, when a radial load is applied to the fixed ring 11, the load is transmitted from the fixed ring 11 to the rotating ring 12 via the rolling elements 13. Within the rolling bearing 10, several rolling elements 13 share the load. The black arrows in Figure 6 indicate the load (rolling element load) received by each rolling element 13. The magnitude of the load received by each rolling element 13 differs depending on the position of the rolling element 13. The "load zone" refers to the range within which the rolling elements 13 receive the load.

[0035] When a separation occurs on the rotating wheel 12, several rolling elements 13 in the load zone enter the separation in sequence and then escape from the separation, causing a pair of positive and negative peaks to appear at the pitch intervals of the rolling elements 13 in the displacement vibration waveform.

[0036] Here, the number of peeling passages N can be calculated by counting the number of peaks (for example, peaks with positive values) that occur when the rolling element 13 enters the peeling, which are included in the fluctuating vibration waveform during the rotation period of the rotating wheel 12. Note that the number of peeling passages N can also be calculated by counting the number of peaks (for example, peaks with negative values) that occur when the rolling element 13 escapes from the peeling, which are included in the fluctuating vibration waveform during the rotation period of the rotating wheel 12.

[0037] Then, based on the calculated number of peeling passes N, it can be estimated whether the peeling has occurred on the fixed wheel 11 or the rotating wheel 12. Specifically, if the number of peeling passes N is equal to the number of times the rolling element 13 passes one observation point on the fixed wheel 11 (for example, the lowest point of the fixed wheel 11) during one rotation of the rotating wheel 12, it can be estimated that the peeling has occurred on the fixed wheel 11 (see FIG. 4).

[0038] On the other hand, if the number of times N that the flaking passes is equal to the number of times that one observation point on the rotating ring 12 comes into contact with the rolling element 13 in the load zone of the rolling bearing 10 (see Figure 6), it can be assumed that flaking has occurred on the rotating ring 12 (see Figure 5).

[0039] Furthermore, by applying a similar method, it can also be estimated that one flaking has occurred on the rolling element 13. Specifically, if the number of flaking passes N is equal to the number of times that one observation point on one rolling element 13 comes into contact with the fixed ring 11 and the rotating ring 12 in the load zone of the rolling bearing 10 (see FIG. 6) during one rotation of the rotating ring 12, it can be estimated that flaking has occurred on the rolling element 13.

[0040] In this specification, "the number of times is equal" includes a case where the number of times is completely the same, and a case where the number of times is substantially the same with some discrepancy.

[0041] In this way, it is possible to estimate where the peeling has occurred on the fixed wheel 11, the rotating wheel 12, and the rolling element 13 from the number of peeling passes N. The threshold values ​​for estimating the location of the peeling are given by the following equations (1) to (3). Equation (1) represents the theoretical value of the number of peeling passes N when peeling has occurred on the fixed wheel 11. Equation (2) represents the theoretical value of the number of peeling passes N when peeling has occurred on the rotating wheel 12. Equation (3) represents the theoretical value of the number of peeling passes N when peeling has occurred on the rolling element 13.

[0042]

number

[0043] Here, Z is the number of rolling elements 13, dp is the pitch circle diameter of the rolling elements 13, dw is the diameter of the rolling elements 13, α is the contact angle, and θ is the angle of the load zone. These specification data are stored in advance in the storage device 105 (see FIG. 2) in association with the bearing model number.

[0044] The theoretical value of the number of peeling passes N when peeling occurs on the fixed wheel 11 corresponds to an example of the "first threshold value," the theoretical value of the number of peeling passes N when peeling occurs on the rotating wheel 12 corresponds to an example of the "second threshold value," and the theoretical value of the number of peeling passes N when peeling occurs on the rolling body 13 corresponds to an example of the "third threshold value."

[0045] The location of the peeling can be estimated by determining whether the number of times N that the peeling passes, extracted from the displacement vibration waveform for one rotation, is equal to one of the three threshold values ​​given by equations (1) to (3).

[0046] (3) Variation index V Before explaining the variation index V, we will explain "pitch interval flaking," which is one type of damage to the rolling bearing 10. Fig. 7 is a schematic diagram for explaining pitch interval flaking. Fig. 7 shows a case where pitch interval flaking has occurred on the raceway surface of the rotating ring 12.

[0047] If the rolling bearing 10 continues to operate after one flaking has occurred on the raceway surface of the rotating ring 12, the damage may progress and multiple flaking may occur at the pitch intervals of the rolling elements 13, as shown in Figure 7. In this specification, multiple flaking that occurs at the pitch intervals of the rolling elements 13 is referred to as "pitch interval flaking." Pitch interval flaking can also occur on the fixed ring 11.

[0048] 8 and 9 are diagrams showing examples of displacement vibration waveforms over the time it takes for the rotating ring 12 to make four rotations. Fig. 8 shows a displacement vibration waveform when pitch interval flaking has occurred on the raceway surface of the fixed ring 11. Fig. 9 shows a displacement vibration waveform when pitch interval flaking has occurred on the raceway surface of the rotating ring 12. The vertical axis of Figs. 8 and 9 shows vibration displacement, and the horizontal axis shows the number of rotations of the rotating ring 12.

[0049] As shown in Figure 8, when pitch interval flakes occur on the fixed wheel 11, a pair of positive and negative peaks appear in the displacement vibration waveform at the pitch intervals of the rolling elements 13, just as when one flake occurs on the fixed wheel 11 (see Figure 4). Therefore, the number of flake passages N when pitch interval flakes occur on the fixed wheel 11 is equal to the number of flake passages N when one flake occurs on the fixed wheel 11.

[0050] However, when attention is paid to the magnitude (amplitude) of the peak when the rolling element 13 enters the flakes, when pitch interval flakes have occurred on the fixed ring 11, there is variation in the magnitude among the multiple peaks, which differs from the case when a single flake has occurred on the fixed ring 11. This variation occurs because the timing at which the rolling element 13 enters the multiple flakes differs due to factors such as differences in size (circumferential width, etc.) of the multiple flakes occurring on the fixed ring 11.

[0051] As shown in Fig. 9, when pitch interval flakes occur on the rotating ring 12, similar to the case where one flake occurs on the rotating ring 12 (see Fig. 5), for each rotation of the rotating ring 12, during the period when the rotating ring 12 enters the load zone, a pair of positive and negative peaks appear in the displacement vibration waveform at the pitch intervals of the rolling elements 13. Therefore, the number of flake passages N when pitch interval flakes occur on the rotating ring 12 is equal to the number of flake passages N when one flake occurs on the rotating ring 12.

[0052] Furthermore, similar to the case where one flake has occurred on the rotating wheel 12 (see Figure 5), in the displacement vibration waveform for each rotation, multiple positive peaks that occur when the rolling element 13 enters the flake are amplitude modulated, and multiple negative peaks that occur when the rolling element 13 escapes from the flake are amplitude modulated.

[0053] Here, from the displacement vibration waveform shown in FIG. 9, the maximum value of the peak (e.g., positive peak) when the rolling element 13 enters the peeling is extracted for each rotation, and the maximum values ​​of the peaks over four rotations are compared. The four circles on the positive side in FIG. 9 indicate the maximum value of the positive peak for each rotation. From FIG. 9, it can be seen that the maximum value of the positive peak for four rotations varies. Furthermore, the four circles on the negative side in FIG. 9 indicate the maximum value of the peak (e.g., negative peak) when the rolling element 13 escapes the peeling for each rotation. It can also be seen that the maximum value of the negative peak for four rotations varies. This variation in the maximum value of the peaks is due to differences in the size (e.g., width in the circumferential direction) of the multiple peelings occurring on the rotating ring 12, which causes differences in the timing at which the rolling element 13 enters the multiple peelings in the load zone.

[0054] In this embodiment, the degree of variation in amplitude in the displacement vibration waveform is extracted as a "variation index V." The variation index V can be obtained by the following procedure.

[0055] First, the maximum positive peak value ai is extracted for each rotation from the displacement vibration waveform for n rotations (n ​​is an integer equal to or greater than 2) (i is the number of rotations, 1≦i≦n). Next, the average value am of the extracted maximum positive peak values ​​a1 to an for n rotations is calculated. FIG. 10 shows a displacement vibration waveform for two rotations when pitch interval peeling occurs in the fixed wheel 11 (n=2). The feature extraction unit 50 extracts the maximum positive peak value a1 for the first rotation and the maximum positive peak value a2 for the second rotation from the displacement vibration waveform for two rotations, and calculates the average value am of the two extracted maximum values ​​a1 and a2 (am=(a1+a2) / 2).

[0056] 11 shows displacement vibration waveforms for two rotations (n=2) when pitch interval separation occurs on the rotating wheel 12. The feature extraction unit 50 extracts the maximum positive peak value a1 of the first rotation and the maximum positive peak value a2 of the second rotation from the displacement vibration waveforms for two rotations, and calculates the average value am of the two extracted maximum values ​​a1 and a2 (am=(a1+a2) / 2).

[0057] Finally, the feature extraction unit 50 calculates the variation index V by substituting the number of rotations n of the displacement vibration waveform to be evaluated, the maximum values ​​a1 to an of the positive peaks per rotation obtained from the displacement vibration waveform, and their average value am into the following equation (4).

[0058]

number

[0059] The variation index V increases as the variation in the maximum peak value ai per rotation increases. Therefore, it is possible to determine whether a single peeling or pitch interval peeling has occurred from the magnitude of the variation index V. Specifically, a threshold value Vth for determining whether a pitch interval peeling has occurred is set in advance through experiments or simulation, and by comparing the variation index V extracted from the displacement vibration waveform with the threshold value Vth, it is possible to determine whether a single peeling or pitch interval peeling has occurred. The method for setting the threshold value Vth will be explained in detail later. The threshold value Vth corresponds to one example of the "fourth threshold value."

[0060] The feature extraction unit 50 extracts a plurality of feature amounts (peel length D, peel pass number N, and variation index V) from the displacement vibration waveform, and outputs the extracted plurality of feature amounts to the damage state diagnosis unit 60.

[0061] (Damage state diagnosis unit 60) The damage state diagnosis unit 60 diagnoses the damage state of the rolling bearing 10 based on a plurality of feature amounts (flaking length D, number of flake passes N, and variation index V) provided by the feature amount extraction unit 50.

[0062] FIG. 12 is a flowchart illustrating the flow of the damage state diagnosis process in the damage state diagnosis unit 60. As shown in FIG. 12, the damage state diagnosis unit 60 first determines in step (hereinafter, step will be abbreviated as "S") 01 whether or not the number of spalling passes N has been extracted from the displacement vibration waveform. If the number of spalling passes N is 0, a NO determination is made in S01. In this case, the damage state diagnosis unit 60 estimates that no spalling has occurred in any of the fixed ring 11, the rotating ring 12, and the rolling elements 13. Therefore, the damage state diagnosis unit 60 diagnoses in S15 that the rolling bearing 10 is normal (without damage).

[0063] On the other hand, when the number of peeling passes N>0, the determination in S01 is YES. In this case, the damage state diagnosis unit 60 estimates that peeling has occurred in any of the fixed wheel 11, the rotating wheel 12, and the rolling element 13. Therefore, the damage state diagnosis unit 60 estimates the location of peeling and the number of peelings based on the multiple feature amounts D, N, and V.

[0064] Specifically, in S02, the damage state diagnosis unit 60 estimates the location of the occurrence of peeling based on the number of peeling passes N. In S02, the number of peeling passes N is compared with a threshold value given by equation (3). This threshold value represents the theoretical value of the number of peeling passes N when peeling occurs on the rolling element 13. If the number of peeling passes N is equal to the threshold value given by equation (3), the determination in S02 is YES. In this case, the damage state diagnosis unit 60 diagnoses in S14 that peeling has occurred on the rolling element 13 (i.e., that the rolling element 13 is damaged).

[0065] On the other hand, if the number of peeling passes N does not match the threshold value given by equation (3) (NO in S02), the damage state diagnosis unit 60 estimates that peeling has occurred in either the fixed wheel 11 or the rotating wheel 12. Next, the damage state diagnosis unit 60 proceeds to S03 and estimates the number of peelings.

[0066] In S03, the damage state diagnosis unit 60 compares the variation index V with a threshold value Vth. If the variation index V is greater than the threshold value Vth (YES in S03), the damage state diagnosis unit 60 estimates that pitch separation has occurred. On the other hand, if the variation index V is equal to or less than the threshold value Vth (NO in S03), the damage state diagnosis unit 60 estimates that one separation has occurred.

[0067] If it is estimated that pitch interval peeling has occurred (YES in S03), the damage state diagnosis unit 60 proceeds to S11 and estimates the location of the pitch interval peeling. In S11, the number of peeling passes N is compared with the threshold values ​​given by equations (1) and (2). The threshold value given by equation (1) represents the theoretical value of the number of peeling passes N when peeling has occurred on the fixed wheel 11. The threshold value given by equation (2) represents the theoretical value of the number of peeling passes N when peeling has occurred on the rotating wheel 12.

[0068] If the number of times N of peeling passes is equal to the threshold value given by equation (1) in S11, the damage state diagnosis unit 60 diagnoses in S12 that the fixed wheel 11 is damaged and that pitch interval peeling has occurred on the fixed wheel 11. If the number of times N of peeling passes is equal to the threshold value given by equation (2) in S11, the damage state diagnosis unit 60 diagnoses in S13 that the rotating wheel 12 is damaged and that pitch interval peeling has occurred on the rotating wheel 12.

[0069] Returning to S03, if it is estimated that one peel has occurred (NO in S03), the damage state diagnosis unit 60 proceeds to S04 and estimates the number of passing rolling elements. In S04, the peel length D is compared with the pitch interval of the rolling elements 13. If the peel length D is equal to or less than the pitch interval of the rolling elements 13 (NO in S04), the damage state diagnosis unit 60 estimates that the number of passing rolling elements is 0 to 1. Then, the damage state diagnosis unit 60 proceeds to S05 and estimates the location where one peel has occurred. In S05, the number of peel passages N is compared with the threshold given by equations (1) and (2). If the number of peel passages N is equal to the threshold given by equation (1) in S05, the damage state diagnosis unit 60 diagnoses in S06 that the fixed wheel 11 is damaged and that one peel has occurred on the fixed wheel 11. Furthermore, it diagnoses that the number of passing rolling elements in the peel is 0 to 1.

[0070] If the number of times N of passing flaking is equal to the threshold value given by equation (2) in S05, the damage state diagnosis unit 60 diagnoses in S07 that the rotating wheel 12 is damaged and that one flaking has occurred on the rotating wheel 12. It also diagnoses that the number of passing rolling elements in the flaking is 0 to 1.

[0071] Returning to S04, if the peel length D is greater than the pitch interval of the rolling elements 13 (YES in S04), the damage state diagnosis unit 60 estimates that the number of passing rolling elements is 1 to 2. Then, the damage state diagnosis unit 60 proceeds to S08 and estimates the location where one peel has occurred. In S08, the number of peel passes N is compared with the threshold given by equations (1) and (2). If the number of peel passes N is equal to the threshold given by equation (1) in S08, the damage state diagnosis unit 60 diagnoses in S09 that the fixed wheel 11 is damaged and that one peel has occurred on the fixed wheel 11. It also diagnoses that the number of passing rolling elements in that peel is 1 to 2.

[0072] If the number of times N of passing flaking is equal to the threshold value given by equation (2) in S05, the damage state diagnosis unit 60 diagnoses in S10 that the rotating wheel 12 is damaged and that one flaking has occurred on the rotating wheel 12. It also diagnoses that the number of passing rolling elements in the flaking is 1 to 2.

[0073] The damage state diagnosis unit 60 outputs damage state data indicating the diagnosis result. In one aspect, the damage state data is transmitted to an external monitoring terminal via the I / F device 104 (see FIG. 2). Based on the estimation result displayed on the monitoring terminal, the operator can determine whether to continue operating the machinery or equipment.

[0074] In Figure 12, the configuration is such that it is estimated whether there is a single peel or a pitch-interval peel based on the variation index V, and then it is estimated based on the number of peel passes N whether the peel has occurred on the fixed wheel 11 or the rotating wheel 12.However, it is also possible to first estimate whether the peel has occurred on the fixed wheel 11 or the rotating wheel 12 based on the number of peel passes N.

[0075] <Effects> As described above, according to this embodiment, a plurality of feature quantities (flaking length D, number of flake passes N, variation index V) are extracted from the vibration waveform of the rolling bearing 10, and each of the extracted feature quantities is compared with a threshold value set for each feature quantity based on the specification data of the rolling bearing 10, thereby making it possible to diagnose the damage state of the rolling bearing 10 with high accuracy.

[0076] Specifically, by comparing the number of flake passages N with the first to third threshold values ​​given by equations (1) to (3) (theoretical values ​​of the number of flake passages N when flakes occur on each of the fixed ring 11, the rotating ring 12, and the rolling elements 13), it is possible to estimate the location of flake occurrence in the rolling bearing 10. Also, by comparing the variation index V with the threshold value Vth (fourth threshold value), it is possible to estimate whether a single flake has occurred in the rolling bearing 10 or whether pitch interval flakes have occurred. Furthermore, by comparing the flake length D with the pitch interval of the rolling elements 13, it is possible to estimate the number of passing rolling elements. This makes it possible to quantitatively and accurately diagnose the damage state, which indicates the degree of damage progression from the occurrence of a single flake in the rolling bearing 10 to the occurrence of pitch interval flakes.

[0077] Example 1 Hereinafter, the damage state diagnosis system according to the present embodiment will be described in further detail with reference to Example 1.

[0078] Figure 13 shows the configuration and operating conditions of the rolling bearing used in Example 1. As shown in Figure 13, the rolling bearing 10 is a roller bearing, and the pitch interval of the rolling elements 13 is 28 mm. The rotation speed of the rotating ring 12 in the rolling bearing 10 was 750 revolutions per minute. When a radial load of 90 kN was applied to the rolling bearing 10, the number of times that one observation point on the rotating ring 12 came into contact with the rolling elements 13 in the load zone of the rolling bearing 10 was 3.

[0079] In Example 1, spalling was formed on the raceway surface of the rotating ring 12. The spalling conditions are shown in Figure 14. As shown in Figure 14, the spalling conditions include a normal condition and four damage conditions that are set in stages according to the progression of spalling.

[0080] The normal condition indicates a state in which there is no peeling on the rotating wheel 12. Damage condition 1 indicates a state in which there is one peeling and the size of the peeling in the circumferential direction is 4 mm. Damage condition 2 indicates a state in which there is one peeling and the size of the peeling in the circumferential direction is 8 mm. Damage condition 3 indicates a state in which there is one peeling and the size of the peeling in the circumferential direction is 18 mm. Damage condition 4 indicates a state in which there are three peelings and the size of each peeling in the circumferential direction is 18 mm. In other words, damage condition 4 indicates a state in which pitch interval peeling has occurred.

[0081] The true values ​​of the damage state under each condition are shown in the rightmost column of Fig. 14. The true values ​​include the correct information on the location of spalling, the number of spalls (one spall / pitch interval spall), and the number of passing rolling elements.

[0082] In Example 1, for each of the five conditions shown in Fig. 14, a displacement vibration waveform of the rolling bearing 10 was generated from the output signal of the sensor 20, and multiple feature quantities (flaking length D, number of flake passes N, and variation index V) were extracted from the generated displacement vibration waveform. Figs. 15 to 17 show the feature quantities extracted under each condition.

[0083] FIG. 15 is a diagram showing the number of flake passages N extracted from the displacement vibration waveform under each condition. As shown in FIG. 15, the number of flake passages N under normal conditions (no flakes) is 0. The number of flake passages N under damage conditions 1 to 4 is 3. The number of flake passages N under these damage conditions 1 to 4 matches the number of times that one observation point on the rotating ring 12 comes into contact with the rolling element 13 in the load zone of the rolling bearing 10. Therefore, it can be seen that the location where the flakes have occurred can be estimated to be the rotating ring 12 from the number of flake passages N.

[0084] Fig. 16 is a diagram showing the variation index V extracted from the displacement vibration waveform under each condition. As shown in Fig. 16, the variation index V under damage condition 4 is more than twice as large as the variation index V under normal condition and damage conditions 1 to 3. Fig. 16 shows that the variation index V appropriately represents the difference in amplitude variation in the displacement vibration waveform between the normal condition and damage conditions 1 to 3 (no peeling or one peeling) and damage condition 4 (pitch-interval peeling).

[0085] Fig. 17 shows the delamination length D extracted from the displacement vibration waveform under each condition. In Fig. 17, black circles indicate the delamination length D extracted from the displacement vibration waveform, and white circles indicate the true value of the delamination length shown in Fig. 14. Under all of damage conditions 1 to 4, the delamination length D shows good agreement with the true value.

[0086] As described above, the multiple feature quantities extracted from the displacement vibration waveform (flaking length D, flake pass count N, and variation index V) accurately represent the damage state of the rolling bearing 10. Therefore, it is possible to quantitatively and accurately diagnose the damage state of the rolling bearing 10 based on the multiple feature quantities.

[0087] <Setting the threshold Vth> A method for setting the threshold value Vth for the variation index V will be described below.

[0088] As shown in Figure 16, the variation index V accurately represents the difference in amplitude variation in the displacement vibration waveform depending on the number of peels. Therefore, by appropriately setting the threshold value Vth and comparing the variation index V with the threshold value Vth, it is possible to estimate whether a single peel or multiple peels have occurred.

[0089] In this embodiment, the threshold value Vth is set based on the static displacement when a radial load is applied to the rolling bearing 10 with one rolling element 13 entering a flake. Figure 18 is a diagram showing the change in the ratio of static displacement with respect to the number of flakes. In Figure 18, when one flake has occurred on the fixed ring 11 or the rotating ring 12, the static displacement when a radial load is applied with one rolling element 13 entering the flake is set to a reference value (1.0). The static displacement ratio represents the ratio of static displacement to the reference value when the number of flakes increases to two or more. Note that two or more flakes correspond to "pitch interval flakes."

[0090] 18, the static displacement ratio increases by approximately two times as the number of flakes increases from 1 to 2. However, as mentioned above, when pitch-spaced flakes occur, the timing at which the rolling element 13 enters the multiple flakes differs, resulting in variations in the amplitude when the rolling element 13 enters the flakes.

[0091] Therefore, in this embodiment, when the number of peelings is 2, it is assumed that the ratio between the minimum value amin=min{a1-an} of the maximum values ​​a1-an of the peaks for n rotations and the maximum value amax=max{a1-an} of the maximum values ​​a1-an falls within the range of 1 to 2. Furthermore, it is assumed that the maximum value amax is 1.5, the median value of this range, the minimum value amin is 1.0, and the maximum values ​​a1-an of the peaks for n rotations excluding the maximum value amax and the minimum value amin are linearly interpolated using the maximum value amx and the minimum value amin. Under these assumptions, the variation index V of the displacement vibration waveform for n rotations is calculated from equation (4), and the calculated variation index V is set as the threshold value Vth.

[0092] FIG. 19 shows the results of calculating the variation index V when n=4. The maximum values ​​a1 to a4 of the peaks over four rotations shown in FIG. 19 are values ​​set based on the assumption of the maximum values ​​a1 to an of the peaks over n rotations described above. By substituting the maximum values ​​a1 to a4 into equation (4), the variation index V=0.172 was calculated. Therefore, the threshold value Vth is set to 0.172, and if the variation index V extracted from the displacement vibration waveform is 0.172 or greater, it is assumed that pitch separation has occurred.

[0093] In the above-described example, the variation index V is smaller than 0.172 under damage conditions 1 to 3 (the number of peels is 1), and is larger than 0.172 under damage condition 4 (the number of peels is 3), as shown in Fig. 16. This suggests that setting the threshold value Vth to 0.172 is appropriate.

[0094] <Other configuration examples> (1) Correction of the variability index V In the above-described example, the variation index V under normal conditions shows a higher value than the variation index V under damage conditions 1 to 3 (see FIG. 16). This is because the vibration of the rolling bearing 10 becomes more irregular when there is no flaking, compared to when there is flaking. The phenomenon of the variation index V being larger under normal conditions can be improved by modifying equation (4) for deriving the variation index V.

[0095] 20 and 21 are diagrams showing examples of displacement vibration waveforms over the time it takes for the rotating wheel 12 to make four rotations. Fig. 20 shows a displacement vibration waveform under normal conditions (no spalling). Fig. 21 shows a displacement vibration waveform under damage condition 4 (pitch interval spalling). The vertical axis of Fig. 20 and Fig. 21 shows the vibration displacement, and the horizontal axis shows the number of rotations of the rotating wheel 12.

[0096] 20, irregular peaks unrelated to the pitch interval of the rolling elements 13 appear in the displacement vibration waveform under normal conditions. Therefore, the feature extraction unit 50 extracts maximum peak values ​​a1 to a4 unrelated to damage from the displacement vibration waveform for each rotation. Therefore, the time intervals t1, t2, and t3 at which the maximum values ​​of the peaks for each rotation appear do not coincide with the rotation period T of the rotating wheel 12.

[0097] In contrast, as shown in FIG. 21, when pitch interval flaking occurs, the time intervals t1, t2, and t3 at which the maximum values ​​a1 to a4 of the peaks per rotation appear approximately coincide with the rotation period T of the rotating wheel 12.

[0098] Therefore, the derivation formula (4) of the variation index V is modified using the time interval ti at which the maximum value ai of the peaks per rotation appears and the rotation period T of the rotating wheel 12. The modified variation index Vr is given by formula (5).

[0099]

number

[0100] Here, ti represents the time interval between the timing at which the maximum peak value ai appears in the displacement vibration waveform for the i-th rotation and the timing at which the maximum peak value ai+1 appears in the displacement vibration waveform for the (i+1)-th rotation, T represents the rotation period of the rotating wheel 12, and k represents an adjustment coefficient.

[0101] Fig. 22 is a diagram showing the variation index Vr calculated using equation (5). Fig. 22 shows the variation index Vr under normal conditions and damage conditions 1 to 4 of the above-mentioned example. The adjustment coefficient k in equation (5) was set to 3, and the variation index Vr was calculated from the displacement vibration waveform under each condition.

[0102] As shown in FIG. 22, the variation index Vr under normal conditions is smaller than the variation index Vr under damage conditions 1 to 4. Also, as in FIG. 16, the variation index Vr under damage condition 4 is higher than the variation index Vr under other conditions. In FIG. 22 as well, the variation index Vr under normal conditions and damage conditions 1 to 3 is less than the threshold value Vth (0.172), and the variation index Vr under damage condition 4 is larger than Vth. Therefore, the corrected variation index Vr can also be used to estimate whether a single peeling or pitch-interval peeling has occurred.

[0103] The adjustment coefficient k in equation (5) may take any value, and may be adjusted, for example, so that the variation index Vr under normal conditions is less than the threshold value Vth.

[0104] (2) Modified example of damage condition diagnosis system Fig. 23 is a block diagram showing a modified example of the damage state diagnosis system according to the present embodiment. As shown in Fig. 23, the damage state diagnosis system according to this modified example differs from the damage state diagnosis system shown in Fig. 1 in the configuration of diagnosis device 100.

[0105] 1 in that it includes a damage state acquisition unit 70, a data acquisition unit 80, a model generation unit 90, and a trained model 110. The diagnostic device 100 according to this modification is configured to apply the trained model 110 to a vibration waveform (e.g., a displacement vibration waveform) that indicates vibrations that occur when the rolling bearing 10 is rotated, and to diagnose the damage state of the rolling bearing 10.

[0106] Specifically, the pre-processing unit 30 generates a vibration waveform (for example, a displacement vibration waveform) from the output signal of the sensor 20 while the rolling bearing 10M to be learned is rotating. The pre-processing unit 30 stores the generated vibration waveform in the storage unit 40. The feature extraction unit 50 extracts a plurality of feature quantities (flaking length D, number of flaking passes N, and variation index V) from the vibration waveform of the rolling bearing 10M stored in the storage unit 40.

[0107] The damage state acquisition unit 70 acquires damage state data indicating the damage state of the rolling bearing 10M. The data acquisition unit 80 acquires multiple feature quantities (flaking length D, number of flake passes N, and variation index V) of the vibration waveform of the rolling bearing 10M extracted by the feature quantity extraction unit 50, and the damage state data of the rolling bearing 10M. The dashed lines in Figure 23 represent the flow of data during learning.

[0108] The model generation unit 90 uses the multiple feature amounts and damage state data acquired by the data acquisition unit 80 as training data to generate a trained model 110 for diagnosing the damage state of the rolling bearing 10. The training method used by the model generation unit 90 to generate the trained model 110 is not particularly limited as long as it is machine learning. For example, a convolutional neural network (CNN) can be used for machine learning.

[0109] The damage state diagnosis unit 60 uses the generated trained model 110 to diagnose the damage state of the rolling bearing 10. Specifically, the damage state diagnosis unit 60 diagnoses the damage state of the rolling bearing 10 by inputting a plurality of feature quantities (peel length D, number of peel passes N, and variation index V) provided by the feature quantity extraction unit 50 into the trained model 110.

[0110] (3) Modification of the feature extraction unit 50 In the above-described embodiment, the feature extraction unit 50 is configured to extract, as one of a plurality of feature quantities, the flaking length D in the circumferential direction of the rolling bearing 10 from the vibration waveform stored in the storage unit 40. The feature extraction unit 50 may also be configured to extract, instead of the flaking length D, the non-flaking length Dn in the circumferential direction of the rolling bearing 10.

[0111] In this modification, a method for extracting the non-peeled length Dn and a process for diagnosing the damaged state using the non-peeled length Dn will be described.

[0112] Fig. 24 is a diagram showing an example of a displacement vibration waveform, and is a diagram to be compared with Fig. 3. Fig. 24 shows a displacement vibration waveform when one flaking occurs on the raceway surface of fixed ring 11.

[0113] In the example of Fig. 24, when the rolling element 13 enters a peeling on the fixed wheel 11, the displacement vibration waveform has a positive peak, and when the rolling element 13 escapes from the peeling, the displacement vibration waveform has a negative peak. In the displacement vibration waveform, a pair of positive and negative peaks accompanying the entry and exit of the rolling element 13 into and from the peeling appear at pitch intervals of the rolling element 13. The amplitude threshold for detecting entry into and exit from the peeling can be determined, for example, from the quartile of the amplitude of the vibration waveform or a multiple of the effective value of the normal value.

[0114] If the operating speed of the rolling bearing 10 is constant, the non-flaking length Dn can be estimated based on the time difference between the time when the rolling element 13 enters the flake and the time when the rolling element 13 exits the flake, and the pitch interval of the rolling elements 13. The non-flaking length Dn corresponds to the difference obtained by subtracting the flake length D from the pitch interval of the rolling elements 13. Therefore, just like the flake length D, the number of rolling elements 13 passing through the flake (the number of passing rolling elements) can also be estimated from the estimated non-flaking length Dn.

[0115] Specifically, as shown in FIG. 24, when the peeling length D is equal to or less than the pitch interval of the rolling elements 13, the non-peeled length Dn is 0 or greater. When the non-peeled length Dn is 0 or greater, the number of passing rolling elements is 0 to 1. Although not shown, when the peeling length D exceeds the pitch interval of the rolling elements 13, the non-peeled length Dn becomes less than 0. When the non-peeled length Dn is less than 0, the number of passing rolling elements is 1 to 2, and there is always the possibility of a rolling element 13 not bearing a load. By calculating the non-peeled length Dn in this way, the number of passing rolling elements can be estimated.

[0116] In this modified example, the feature extraction unit 50 extracts a plurality of feature amounts (non-peeling length Dn, number of peeling passes N, and variation index V) from the displacement vibration waveform, and outputs the extracted plurality of feature amounts to the damage state diagnosis unit 60. The damage state diagnosis unit 60 diagnoses the damage state of the rolling bearing 10 based on these plurality of feature amounts (non-peeling length Dn, number of peeling passes N, and variation index V).

[0117] Fig. 25 is a flowchart illustrating the flow of damage state diagnosis processing in the damage state diagnosis unit 60 according to this modification. The flowchart shown in Fig. 25 differs from the flowchart shown in Fig. 12 in that S04 is replaced with S04A.

[0118] As shown in FIG. 25, if it is estimated in S03 that one peel has occurred (NO in S03), the damage state diagnosis unit 60 proceeds to S04A, where it estimates the number of passing rolling elements. In S04A, the non-peeling length Dn is determined based on 0. If the non-peeling length Dn is 0 or more (NO in S04A), the damage state diagnosis unit 60 estimates that the number of passing rolling elements is 0 to 1. Then, the damage state diagnosis unit 60 proceeds to S05, where it estimates the location where one peel has occurred.

[0119] In S04A, if the non-peeling length Dn is smaller than 0 (YES in S04A), the damage state diagnosis unit 60 estimates that the number of passing rolling elements is 1 to 2. Then, the damage state diagnosis unit 60 proceeds to S08, where it estimates the location where one peeling has occurred.

[0120] In this modified example, multiple feature quantities (non-peeling length Dn, number of peeling passes N, variation index V) are extracted from the vibration waveform of the rolling bearing 10, and each of the multiple extracted feature quantities is compared with a threshold value set for each feature quantity based on the specification data of the rolling bearing 10, thereby making it possible to estimate the damage state of the rolling bearing 10 with high accuracy.

[0121] <Example 2> A damage state diagnosis system according to this modification will be described in further detail below with reference to a second embodiment.

[0122] The configuration and operating conditions of the rolling bearing used in Example 2 are the same as those shown in Figure 13 for Example 1. That is, the rolling bearing 10 is a roller bearing, and the pitch interval of the rolling elements 13 is 28 mm. The rotation speed of the rotating ring 12 in the rolling bearing 10 is 750 revolutions per minute. When a radial load of 90 kN is applied to the rolling bearing 10, the number of times that one observation point on the rotating ring 12 comes into contact with the rolling elements 13 in the load zone of the rolling bearing is 3.

[0123] In Example 2, spalling was formed on the raceway surface of the rotating ring 12. The spalling conditions are shown in Figure 26. As shown in Figure 26, the spalling conditions include a normal condition and four damage conditions that are set in stages according to the progression of spalling.

[0124] The normal condition indicates a state in which there is no spalling on the rotating wheel 12, and the size of the non-peeling in the circumferential direction is equal to the pitch interval of 28 mm of the rolling elements 13. Damage condition 1 indicates a state in which there is one spalling and the size of the non-peeling in the circumferential direction is 24 mm. Damage condition 2 indicates a state in which there is one spalling and the size of the non-peeling in the circumferential direction is 20 mm. Damage condition 3 indicates a state in which there is one spalling and the size of the non-peeling in the circumferential direction is 10 mm. Damage condition 4 indicates a state in which there are three spallings and the size of each non-peeling in the circumferential direction is 10 mm. In other words, damage condition 4 indicates a state in which pitch interval spalling has occurred.

[0125] The true values ​​of the damage state under each condition are shown in the rightmost column of Fig. 26. The true values ​​include the correct information on the location of spalling, the number of spalls (one spall / pitch interval spall), and the number of passing rolling elements.

[0126] In Example 2, for each of the five conditions shown in Fig. 26, a displacement vibration waveform of the rolling bearing 10 was generated from the output signal of the sensor 20, and multiple feature quantities (non-peeling length Dn, number of peeling passes N, and variation index V) were extracted from the generated displacement vibration waveform. Figs. 27 to 29 show the feature quantities extracted under each condition.

[0127] Fig. 27 is a diagram showing the number of flake passages N extracted from the displacement vibration waveform under each condition. As shown in Fig. 27, the number of flake passages N under normal conditions (no flakes) is 0. The number of flake passages N under damage conditions 1 to 4 is 3. The number of flake passages N under these damage conditions 1 to 4 matches the number of times that one observation point on the rotating ring 12 comes into contact with the rolling element 13 in the load zone of the rolling bearing 10. Therefore, it can be seen that the location where the flakes have occurred can be estimated to be the rotating ring 12 from the number of flake passages N.

[0128] Fig. 28 is a diagram showing the variation index V extracted from the displacement vibration waveform under each condition. As shown in Fig. 28, the variation index V under damage condition 4 is more than twice as large as the variation index V under normal condition and damage conditions 1 to 3. Fig. 28 shows that the variation index V appropriately represents the difference in the variation in amplitude in the displacement vibration waveform between the normal condition and damage conditions 1 to 3 (no peeling or one peeling) and damage condition 4 (pitch-interval peeling).

[0129] Fig. 29 is a diagram showing the non-peeling length Dn extracted from the displacement vibration waveform under each condition. In Fig. 29, black circles indicate the non-peeling length Dn extracted from the displacement vibration waveform, and white circles indicate the true value of the non-peeling length shown in Fig. 26. Under all of damage conditions 1 to 4, the non-peeling length Dn shows good agreement with the true value.

[0130] As described above, the multiple feature amounts extracted from the displacement vibration waveform (non-flaking length Dn, number of flake passes N, and variation index V) accurately represent the damage state of the rolling bearing 10. Therefore, it is possible to quantitatively and accurately diagnose the damage state of the rolling bearing 10 based on the multiple feature amounts.

[0131] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present disclosure is defined by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0132] 10 Rolling bearing, 11 Fixed ring, 12 Rotating ring, 13 Rolling element, 14 Cage, 20 Sensor, 30 Pre-processing unit, 40 Memory unit, 50 Feature extraction unit, 60 Damage state diagnosis unit, 70 Damage state acquisition unit, 80 Data acquisition unit, 90 Model generation unit, 101 CPU, 102 RAM, 103 ROM, 104 I / F unit, 105 Storage device, 106 Communication bus, 110 Trained model.

Claims

1. A damage condition diagnosis system for diagnosing a damage condition indicating the degree of progression of damage to a rolling bearing, wherein the rolling bearing includes a fixed ring, a rotating ring, and a plurality of rolling elements circumferentially arranged at equal pitch intervals between the fixed ring and the rotating ring so as to be able to roll; a sensor for detecting vibrations that occur when the rolling bearing is rotated; a diagnostic device that diagnoses a damaged state of the rolling bearing based on an output signal of the sensor, The diagnostic device comprises: a pre-processing unit that acquires an output signal from the sensor over a period of time during which the rotating wheel rotates multiple times to generate a vibration waveform; a feature extraction unit that extracts a plurality of feature amounts from the vibration waveform; a damage state diagnosis unit that diagnoses a damage state of the rolling bearing based on the plurality of feature amounts extracted by the feature amount extraction unit, The plurality of feature quantities include a flake length or a non-flake length in the circumferential direction of the rolling bearing, a flake pass count that each rolling element passes through a flake within the time it takes for the rotating ring to make one rotation, and a variation index that represents the variation in amplitude of the vibration waveform.

2. 2. The damage state diagnosis system according to claim 1, wherein the damage state diagnosis unit diagnoses the damage state of the rolling bearing by comparing each of the plurality of feature amounts with a threshold value set for each feature amount based on specification data of the rolling bearing.

3. the feature extraction unit acquires, from the vibration waveform, an entry time at which the rolling element enters a flake on a raceway ring and an exit time at which the rolling element exits the flake, and estimates the flake length based on the time difference between the acquired entry time and exit time; 3. The damage state diagnosis system according to claim 2, wherein the damage state diagnosis unit estimates the number of passing rolling elements, which is the number of rolling elements passing through the peeling, by comparing the peeling length estimated by the feature extraction unit with the pitch interval of the plurality of rolling elements.

4. 4. The damage state diagnosis system according to claim 3, wherein the damage state diagnosis unit estimates that the number of passing rolling elements is 0 to 1 when the peeling length estimated by the feature extraction unit is equal to or less than the pitch interval of the plurality of rolling elements, and estimates that the number of passing rolling elements is 1 to 2 when the peeling length estimated by the feature extraction unit is greater than the pitch interval of the plurality of rolling elements.

5. the feature extraction unit acquires, from the vibration waveform, an entry time at which the rolling element enters a flake on a raceway ring and an exit time at which the rolling element exits the flake, and estimates the non-flake length based on the time difference between the acquired entry time and exit time and the pitch interval between the plurality of rolling elements; 3. The damage state diagnosis system according to claim 2, wherein the damage state diagnosis unit estimates the number of passing rolling elements, which is the number of rolling elements passing through the peeling, from the non-peeling length estimated by the feature extraction unit.

6. 6. The damage state diagnosis system according to claim 5, wherein the damage state diagnosis unit estimates that the number of passing rolling elements is 0 to 1 when the non-peeling length estimated by the feature extraction unit is 0 or more, and estimates that the number of passing rolling elements is 1 to 2 when the non-peeling length estimated by the feature extraction unit is less than 0.

7. the feature extraction unit calculates the number of times the rolling element passes through a peel by counting the number of peaks that occur when the rolling element enters a peel or the number of peaks that occur when the rolling element escapes from the peel, the peaks being included in the vibration waveform during a rotation period of the rotating wheel; the damage state diagnosis unit estimates a location where flaking has occurred in the rolling bearing by comparing the number of flaking passes calculated by the feature amount extraction unit with first to third threshold values; 3. The damage state diagnosis system according to claim 2, wherein the first threshold value is a theoretical value of the number of flaking passes when flaking occurs on the fixed ring, the second threshold value is a theoretical value of the number of flaking passes when flaking occurs on the rotating ring, and the third threshold value is a theoretical value of the number of flaking passes when flaking occurs on the rolling element.

8. 8. The damage state diagnosis system according to claim 7, wherein the damage state diagnosis unit estimates that peeling has occurred in the fixed ring when the number of peeling passes is equal to the first threshold value, estimates that peeling has occurred in the rotating ring when the number of peeling passes is equal to the second threshold value, and estimates that peeling has occurred in the rolling element when the number of peeling passes is equal to the third threshold value.

9. 9. The damage state diagnosis system according to claim 7, wherein the first threshold, the second threshold, and the third threshold are set using the number of the plurality of rolling elements, pitch circle diameters of the rolling elements, diameters of the rolling elements, contact angles, and angles of load zones in the rolling bearing.

10. the feature extraction unit extracts from the vibration waveform a maximum value of the peak when the rolling element enters a flaking area for each rotation, calculates an average value of the maximum values ​​of the peaks extracted for a plurality of rotations, and calculates the variation index based on the maximum value and the average value of the peaks for the plurality of rotations; 3. The damage state diagnosis system according to claim 2, wherein the damage state diagnosis unit estimates whether one flaking has occurred in the rolling bearing or whether pitch-interval flaking has occurred in the rolling bearing by comparing the variation index calculated by the feature extraction unit with a fourth threshold value.

11. 11. The damage state diagnosis system according to claim 10, wherein the damage state diagnosis unit estimates that one flaking has occurred in the rolling bearing when the variation index is smaller than the fourth threshold value, and estimates that pitch interval flaking has occurred in the rolling bearing when the variation index is larger than the fourth threshold value.

12. 11. The damage state diagnosis system according to claim 10, wherein the feature extraction unit calculates the variation index based on the maximum value and the average value for the plurality of rotations, a rotation period of the rotating wheel, and a time interval at which the maximum value of the peak appears for each rotation.

13. The damage state diagnosis system according to any one of claims 10 to 12, wherein the fourth threshold value is set based on a ratio of a static displacement when a load is applied to the rolling bearing in a state in which one flaking has occurred on the fixed ring or the rotating ring to the reference value, the static displacement when the number of flaking has changed from 1 to 2.

14. The diagnostic device comprises: a data acquisition unit that acquires the plurality of feature amounts extracted from a vibration waveform generated from an output signal of the sensor while the rolling bearing to be learned is rotating, and damage state data of the rolling bearing to be learned; a model generation unit that generates a trained model for diagnosing a damage state of the rolling bearing using the data acquired by the data acquisition unit as training data, 2. The damage state diagnosis system according to claim 1, wherein the damage state diagnosis unit inputs the plurality of feature quantities extracted by the feature quantity extraction unit into the trained model and diagnoses the damage state of the rolling bearing.

15. A damage condition diagnosis method for diagnosing a damage condition indicating the degree of progression of damage to a rolling bearing, wherein the rolling bearing includes a fixed ring, a rotating ring, and a plurality of rolling elements circumferentially arranged at equal pitch intervals between the fixed ring and the rotating ring so as to be able to roll, detecting, by a sensor, vibrations that occur when the rolling bearing is rotated; and diagnosing a damaged state of the rolling bearing based on an output signal of the sensor, The diagnosing step includes: generating a vibration waveform by acquiring an output signal from the sensor over a period of time during which the rotating wheel rotates multiple times; extracting a plurality of feature amounts from the vibration waveform; and diagnosing a damage state of the rolling bearing based on the extracted plurality of feature amounts, the plurality of feature quantities include a flaking length or a non-flaking length in the circumferential direction of the rolling bearing, a flaking pass count in which each rolling element passes through a flaking area within a time period in which the rotating ring makes one rotation, and a variation index representing a variation in amplitude of the vibration waveform; a damage state diagnosis method, wherein the diagnosing step includes a step of diagnosing the damage state of the rolling bearing by comparing each feature amount with a threshold value set for each feature amount based on specification data of the rolling bearing.

Images