Rotating device state change detection system and rotating device state change detection method

Abstract

This state change detection system comprises: a rotation detection unit that outputs a pulse signal; a pulse calculation unit that obtains a pulse calculation waveform representing a pulse period for each rotation angle of a rotating body; a reference generation unit that generates a reference waveform in which an average pulse calculation waveform, which is obtained by averaging values of the pulse calculation waveform with an accumulated value for each rotation angle, is represented by the distribution for one rotation of the rotating body; a rotational fluctuation waveform generation unit that uses the reference waveform to generate a rotational fluctuation waveform obtained by extracting, from the pulse calculation waveform used to generate the reference waveform or a pulse calculation waveform succeeding the pulse calculation waveform, a fluctuation component other than a steady component that steadily occurs with the rotation of the rotating body; and a state change detection unit that detects a change in rotation state on the basis of the rotational fluctuation waveform.

Description

System for detecting changes in the state of a rotating device and method for detecting changes in the state of a rotating device

[0001] The present invention relates to a system for detecting changes in the state of a rotating device and a method for detecting changes in the state of a rotating device.

[0002] Rotating devices, including couplings that connect rotating bodies such as drive shafts and driven shafts, are widely used for power transmission in various prime movers, electric motors, and speed reducers. For example, in automobiles, couplings are used in hub unit bearings that connect the drive shaft, which is rotated by the driving force from the engine, to the wheel on which the tire is mounted. In rotating devices such as hub unit bearings, the drive shaft and driven shaft each have annular flange portions that extend radially, and power is transmitted by fastening these flange portions together with multiple fasteners such as bolts and nuts.

[0003] For example, the hub unit bearing 500 of Patent Document 1, as shown in Figure 63, comprises an outer ring 501 having a double row of outer ring raceways 501a, a hub 502 having a double row of inner ring raceways 502a, and a plurality of rolling elements 503 that are rotatably provided between the outer ring raceway 501a and the inner ring raceway 502a. Encoders 504 are provided on both axial sides of the hub 502 of the hub unit bearing 500, and sensors (not shown) are provided opposite each encoder 504. Based on the phase difference between the detection signals of these sensors, it is possible to detect the torque applied to the hub 502.

[0004] Japanese Patent Application Publication No. 2004-19934

[0005] In the rotating devices described above, for example, loose fasteners can cause abnormal noise and vibration, poor assembly or aging can cause gaps between flanges, misalignment, or miscoupling. In addition, in the support bearings of the rotating body, damage such as delamination or indentation, or loss of preload may occur on the raceway surface. It is desirable to detect these various changes in the state of rotating devices as early as possible. For example, in the case of damage to the hub unit, the medium carbon steel used as the bearing material has higher toughness than bearing steel (C: 1 mass%), so cracks that occur at the position of the maximum dynamic shear stress do not easily extend toward the raceway surface, and the cracks progress parallel to the raceway surface. In the case of such cracks, in the initial stages the raceway surface does not peel off, and only the part from the crack to the raceway surface is pressed down by the rolling elements and indented, resulting in little deterioration of vibration or acoustics. However, as the crack progresses and the crack appears on the raceway surface, delamination progresses rapidly. Similarly, indentations also affect vibration and acoustics. In cases of failure where delamination progresses rapidly from a certain point in time, early detection of the change in condition is particularly desirable. Furthermore, the damage detection described above is not limited to vehicle hub unit bearings, but is also applicable to other rotating devices, where early detection of changes in condition is desirable.

[0006] Therefore, the present invention aims to provide a system for detecting changes in the state of a rotating device and a method for detecting changes in the state of a rotating device that can reliably and quickly detect changes in the state of the rotating device.

[0007] The present invention comprises the following configuration: (1) A state change detection system for a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: a rotation detection unit that outputs a pulse signal corresponding to the rotation of the rotating body; a pulse calculation unit that obtains a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the pulse signal output from the rotation detection unit; a reference generation unit that generates a reference waveform in which the average pulse calculation waveform obtained by integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body and averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; a rotation fluctuation waveform generation unit that generates a rotation fluctuation waveform by extracting fluctuations other than the steady component that steadily occurs with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or a pulse calculation waveform continuous with the pulse calculation waveform; and a state change detection unit that detects a change in the rotation state based on the rotation fluctuation waveform. (2) A method for detecting a change in the rotation state of a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: outputting a pulse signal corresponding to the rotation of the rotating body; obtaining a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the output pulse signal; integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body; generating a reference waveform in which the average pulse calculation waveform obtained by averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; generating a rotation fluctuation waveform by extracting fluctuations other than the steady component that steadily occurs with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or pulse calculation waveforms continuous with the pulse calculation waveform; and detecting a change in the rotation state based on the rotation fluctuation waveform.

[0008] According to the present invention, changes in the state of a rotating device can be detected reliably and at an early stage.

[0009] Figure 1 is a schematic diagram of the state change detection system for a rotating device. Figure 2 is a schematic diagram showing the flange portions of the first and second rotating bodies, respectively. Figure 3 is a functional block diagram of the control unit. Figure 4 is an explanatory diagram showing how the rotation detection unit detects rotation. Figure 5 is a schematic circuit diagram of the pulse signal generation unit. Figure 6 is an explanatory diagram showing a spatial waveform plotted with a pulse signal corresponding to the sensor output signal and the period of each pulse in this pulse signal. Figure 7 is an explanatory diagram showing an example of another definition of a pulse in a pulse signal. Figure 8 is an explanatory diagram schematically showing the period of an arbitrary pulse in the pulse signal and the pulse for one rotation of the rotating body centered on that pulse. Figure 9 is an explanatory diagram showing an example of a period ratio waveform. Figure 10 is an explanatory diagram schematically showing the process of extracting fluctuations other than the steady-state component that occurs steadily for each rotation angle of the rotating body from the period ratio waveform to generate a rotation fluctuation waveform. Figure 11 is an explanatory diagram showing how the detected period ratio waveform is corrected using a composite average waveform. Figure 12 is a control block diagram showing how to obtain a rotation fluctuation waveform by obtaining a difference waveform from the detected pulse signal. Figure 13 is an explanatory diagram showing time chart 1 from the period ratio waveform to the detection of rotational fluctuation abnormalities. Figure 14 is an explanatory diagram showing time chart 2 from the period ratio waveform to the detection of rotational fluctuation abnormalities. Figure 15 is an explanatory diagram showing time chart 3 from the detection of rotational fluctuation abnormalities from the period ratio waveform. Figure 16 is an explanatory diagram showing the period ratio waveform in the case of a rotational state with misalignment and internal bearing defects. Figure 17 is an explanatory diagram showing the average period ratio waveform of fluctuations caused by internal bearing defects. Figure 18 is an explanatory diagram schematically showing the frequency characteristics of the period ratio waveform in the case of misalignment and internal bearing defects. Figure 19 is an explanatory diagram showing the results of frequency analysis of the waveform after processing the period ratio waveform shown in Figure 16 with bandpass filters BPF1 and BPF2. Figure 20 is an explanatory diagram showing the results of obtaining the frequency characteristics from the difference waveform in the case of a rotational state with misalignment and internal bearing defects. Figure 21 is an explanatory diagram showing the results of applying a low-pass filter to the difference waveform shown in Figure 11. Figure 22 is an explanatory diagram showing the result of applying a high-pass filter to the difference waveform obtained by subtracting the composite average waveform from the period ratio waveform K shown in Figure 16.Figure 23 is a schematic diagram illustrating the difference in anomaly detection accuracy using a reference waveform. Figure 24 is a schematic diagram of a first configuration example of a bearing damage detection system for a hub unit bearing. Figure 25 is a functional block diagram of the control unit. Figure 26 is a schematic diagram illustrating the occurrence of vibration when separation occurs in a part of the outer ring raceway. Figure 27 is a schematic diagram illustrating the change in the sensor output due to the relative displacement between the inner and outer rings caused by the separation of the outer ring raceway. Figure 28 is a schematic diagram illustrating the change in the sensor output signal from the sensor when the rolling element transitions from a floating state to a riding state and back to the original rolling state. Figure 29 is a flowchart illustrating the procedure for determining damage to a hub unit bearing. Figure 30 is a schematic diagram illustrating an example of a waveform from the pulse output signal to the generation of a rotational fluctuation signal by extracting the rotational fluctuation of the hub shaft. Figure 31 is a schematic diagram illustrating an example of a period ratio average value distribution WF6, which is the average value of the period ratio difference for each pole of a magnetic encoder. Figure 32 is a schematic diagram illustrating a period ratio fluctuation waveform WF7. Figure 33 is a schematic diagram showing the frequency characteristics WF8 obtained by FFT processing of the period ratio fluctuation waveform WF7. Figure 34 is a schematic diagram showing the frequency characteristics of the pulse period waveform WF1 shown as a reference example. Figure 35 is a flowchart of another procedure 1 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. Figure 36 is a flowchart of another procedure 2 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. Figure 37 is a flowchart of another procedure 3 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. Figure 38 is a flowchart of another procedure 4 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. Figure 39 is a schematic configuration diagram of a second configuration example of a bearing damage detection system for a hub unit bearing. Figure 40 is a schematic diagram showing the arrangement relationship between the inner ring, outer ring, and magnetic encoder of the hub unit bearing and the sensor detection area DRa and sensor detection area DRb. Figure 41 is a schematic circuit diagram of the pulse signal generation unit. Figure 42 is an explanatory diagram showing examples of waveforms for pulse signals PL_A, PL_B, and phase difference signal PD. Figure 43 is an explanatory diagram showing the effect of the relative displacement of the inner and outer rings due to the separation of the outer ring raceway on the sensor output.Figure 44 is an explanatory diagram showing the relationship between the displacement of the magnetic encoder and the phase difference of the detected pulse signal. Figure 45 is an explanatory diagram showing examples of changes in pulse signals PL_A, PL_B and phase difference signal PD. Figure 46 is an explanatory diagram schematically showing the waveforms of specific pulse signals PL_A, PL_B and phase difference signal PD. Figure 47 is a flowchart showing the procedure for determining damage to the hub unit bearing using the two sensors described above. Figure 48 is an explanatory diagram showing the waveforms of pulse signals PL_A and PL_B. Figure 49 is an explanatory diagram showing the change in period and phase difference, with the horizontal axis representing space representing the order of pulses of pulse signal PL_B and the vertical axis representing time. Figure 50 is an explanatory diagram showing an example of a waveform from the pulse output signal to the generation of a rotation fluctuation signal by extracting the rotation fluctuation of the hub shaft. Figure 51 is an explanatory diagram showing an example of the average value distribution WF16 of the phase difference ratio. Figure 52 is an explanatory diagram schematically showing the phase difference ratio fluctuation waveform WF17. Figure 53 is an explanatory diagram showing an example of a frequency characteristic WF18 obtained by FFT processing of the phase difference ratio fluctuation waveform WF17. Figure 54 is a flowchart of another procedure 1 for damage determination, with some modifications to the steps in the flowchart shown in Figure 47. Figure 55 is a flowchart of another procedure 2 for damage determination, with some modifications to the steps in the flowchart shown in Figure 47. Figure 56 is a flowchart of another procedure 3 for damage determination, with some modifications to the steps in the flowchart shown in Figure 47. Figure 57 is an explanatory diagram showing the relative rotation and displacement between the wheel and the hub when the nut loosens during constant speed rotation, according to the degree of loosening. Figure 58 is an explanatory diagram showing the shape of face-opening between rotating bodies. Figure 59 is an explanatory diagram showing the shape of misalignment between rotating bodies. Figure 60 is a graph showing the distribution of relative changes between the inner and outer rings of a rolling bearing with preload applied. Figure 61 is a graph showing the distribution of relative changes between the inner and outer rings of a rolling bearing with preload removed. Figure 62 is an explanatory diagram showing the waveform of the frequency characteristics for each detection target of the hub unit bearing. Figure 63 is a diagram showing the configuration of a conventional hub unit bearing.

[0010] Embodiments of the present invention will be described in detail below with reference to the drawings. The following description describes a state change detection system for detecting changes in the rotational state of a rotating body in a rotating device having a rotating body and rolling bearings that rotatably support the rotating body. Here, a coupling in which a pair of rotating bodies are connected by a plurality of fasteners is given as an example, but the configuration of the rotating body is not limited to this. <Configuration of the Rotating Device> Figure 1 is a schematic configuration diagram of the state change detection system for the rotating device. The state change detection system 100 comprises a rotating device 11, a rotation detection unit 13, and a control unit 15. The rotating device 11 comprises a first rotating body 17 and a second rotating body 19, a plurality of fasteners 21 connecting the first rotating body 17 and the second rotating body 19, and rolling bearings 23, 25 that support the integrated rotating body 12. The rotating body 12 is rotatably supported on the side of a fixed member such as a housing by the rolling bearings 23, 25. The rotating body 12 shown here is a flange-type coupling, but other configurations are also acceptable as long as power is transmitted between the first rotating body 17 and the second rotating body 19 without the use of elastic members such as rubber, that is, without the use of members or mechanisms that allow deformation to escape.

[0011] The rotation detection unit 13 detects the rotation of the first rotating body 17. The target of rotation detection can be appropriately selected from, for example, the raceway ring on the rotating side of the rolling bearing 23, the rotating body 12, etc. Here, the rotation detection unit 13 detects the rotation of the inner ring 23A of the rolling bearing 23, but if, for example, the rotating body 12 is supported by the outer ring and rotates together with the outer ring, the rotation detection unit 13 detects the rotation of the outer ring.

[0012] The first rotating body 17 has a shaft 27 and an annular flange portion 29 extending radially from one end of the shaft 27. The second rotating body 19 similarly has a shaft 31 and a flange portion 33. The shaft 27 is supported by the inner ring 23A of the rolling bearing 23, and the shaft 31 is supported by the inner ring 25A of the rolling bearing 25. As a result, the first rotating body 17 and the second rotating body 19 are rotatable relative to the fixed member to which the outer rings 23B and 25B are fixed. Multiple through holes 35 are formed in the flange portions 29 and 33 along the circumferential direction, and fasteners 21 are inserted into each through hole 35. The fasteners 21 shown here consist of a bolt 21A and a nut 21B, and the first rotating body 17 and the second rotating body 19 are connected by tightening a pair of flange portions 29 and 33 between the head of the bolt 21A and the nut 21B.

[0013] The control unit 15 detects changes in the rotation state of the rotating device 11 based on the output signal from the rotation detection unit 13. This control unit 15 is configured as a computer equipped with a processor such as a CPU, a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), an SSD (Solid State Drive), and other storage devices.

[0014] Figure 2 is a schematic diagram showing the flange portions 29 and 33 of the first rotating body 17 and the second rotating body 19, respectively. In the following description, the same reference numerals are used for identical members and parts to simplify or omit their explanation.

[0015] A projection shaft 37 is provided at the rotation center of one flange portion 29, projecting in the axial direction. The other flange portion 33 is provided with a recess 39 that accommodates the projection shaft 37. The outer diameter of the projection shaft 37 is slightly smaller than the inner diameter of the recess 39. These pair of flange portions 29 and 33 are fastened together by a plurality of fasteners 21 with the projection shaft 37 inserted into the recess 39 and positioned radially, that is, with their rotational axes aligned by a spigot fit between the projection shaft 37 and the recess 39.

[0016] The rotation detection unit 13 shown in Figure 1 detects the rotation of the first rotating body 17 and outputs a pulse signal corresponding to that rotation. The rotation of the first rotating body 17 can be detected, for example, by a sensor 43 reading the movement of an encoder (not shown) attached to the raceway ring (in this case, the inner ring 23A) on the rotating side of the rolling bearing 23 to which the shaft body 27 is fixed. The output signal from the sensor 43 is converted into a pulse signal corresponding to the rotation of the first rotating body 17 by the pulse signal generation unit 45 and output to the control unit 15. In other words, the rotation detection unit 13 generates pulses at regular rotation angle intervals as the rotating body 12 rotates and outputs a pulse signal corresponding to the rotation of the rotating body 12.

[0017] Figure 3 is a functional block diagram of the control unit 15. The control unit 15 includes a rotational variation extraction unit 47 that determines rotational variation based on the pulse signal described above, and a state change detection unit 55 that detects changes in the rotational state. The rotational variation extraction unit 47 is composed of a pulse calculation unit 49, a reference generation unit 51, and a rotational variation waveform generation unit 53.

[0018] The details of the rotational variation extraction unit 47 will be described later, but in general terms, it operates as follows. The pulse calculation unit 49 calculates pulse values ​​for each rotation angle of the rotating body 12 based on the width and period of each pulse of the pulse signal output from the rotation detection unit 13. The reference generation unit 51 integrates the calculated pulse values ​​for each rotation angle of the rotating body 12. That is, it adds together pulse values ​​generated at the same rotation angle. Then, it calculates an average pulse value by averaging the above integrated values ​​for each rotation angle, for example by dividing the obtained integrated value for each rotation angle by the number of times the integration was performed. The reference waveform is generated by representing the distribution of the average pulse values ​​obtained in this way as one rotation of the rotating body 12.

[0019] The rotational fluctuation waveform generation unit 53 generates a waveform for each rotation angle using the reference waveform generated by the reference generation unit 51, based on the pulse signal output from the rotation detection unit 13. This process extracts fluctuations other than the steady-state components that occur steadily for each rotation angle, and generates the extracted results as a rotational fluctuation waveform. The state change detection unit 55 detects changes in the rotational state based on the generated rotational fluctuation waveform.

[0020] Here, we will briefly explain the example of pulse signal generation and the example of signal control by the control unit 15 described above. Figure 4 is an explanatory diagram showing how the rotation detection unit 13 detects rotation. In Figure 4, the direction of action of the radial load applied from the inner ring 23A to the outer ring 23B of the rolling bearing 23 (generally the vertical direction Z, which is the direction of gravity) is shown as the vertical direction. An annular magnetic encoder 57 is provided on the end face of the inner ring 23A. The detection area DR of the sensor 43 is preferably located in the vertical intermediate area Wa of the magnetic encoder 57, and is particularly preferably located at its center. Here, the rotation detection area of ​​the magnetic encoder 57 by the sensor 43 is schematically shown as a rectangular detection area DR, but the actual size of the detection area DR will be a much smaller area.

[0021] For example, when a radial load is applied from the inner ring 23A to the outer ring 23B, and the rolling elements roll between the outer and inner ring raceways, radial vibrations caused by damage to the raceway surface become more pronounced in the vertical direction Z where the radial load acts. Therefore, detecting rotation within an intermediate region Wa close to the center of the vertical direction Z of the inner ring 23A and outer ring 23B allows for sensitive detection of vertical displacement. The aforementioned intermediate region Wa can be exemplified as the area radially inward of the outer ring raceway surface or the inner ring raceway surface.

[0022] However, the detection area DR by the sensor 43 is not limited to being placed in the intermediate area Wa described above, but may be set at any position such as the upper or lower ends in the vertical direction. In this case, rotation detection is still possible, and the degree of freedom in installing the sensor 43 is not reduced.

[0023] Figure 5 is a schematic circuit diagram of the pulse signal generation unit 45. The pulse signal generation unit 45 outputs the output signal from the sensor 43 as a pulse signal when a drive voltage E adjusted by the resistance value R is applied to the sensor 43. Specifically, if the sensor 43 is a current output type, the current value output from the sensor 43 changes due to the magnetic flux, and a voltage E based on the relationship between the voltage value E, the current value I, and the resistance value R is output as a pulse signal.

[0024] Figure 6 is an explanatory diagram showing the spatial waveform plotted with the pulse signal corresponding to the sensor output signal and the period of each pulse in this pulse signal. Here, an example of a pulse signal when the rotation of the rotating body 12 is detected by a magnetic encoder and a magnetic sensor is shown. The two pulse signals shown in the upper part of Figure 6 represent the magnetic change of the magnetic encoder 57 as it passes through the detection area DR of the sensor 43 in accordance with the rotation of the rotating body 12. One of the pulse signals is an example of bipolar detection, where the signal rises when a magnetic field of the north or south pole is detected, and the other is an example of alternating detection, where the signal rises when a magnetic field of the north or south pole is detected and then falls when a magnetic field of the south or north pole is detected.

[0025] In the case of bipolar detection, when the pole of the magnetic encoder 57 reaches the detection region DR, the sensor output signal changes from a LOW voltage to a HIGH voltage, and the timing of the transition from LOW voltage to HIGH voltage is used as the rising edge timing described above. On the other hand, in the case of alternating detection, when the pole of the magnetic encoder 57 reaches the detection region DR, the sensor output signal changes from a HIGH voltage to a LOW voltage, and the timing of the transition from HIGH voltage to LOW voltage is used as the rising edge timing described above.

[0026] For example, in the case of bipolar detection, if a pair of HIGH voltages from the same type of pole is considered as one pulse, and in the case of alternating detection, if a pair of one HIGH voltage followed by a LOW voltage is considered as one pulse, then if there is an irregularity in the rotation of the first rotating body 17, the pulse period of each pulse (for example, T1 to T7) will fluctuate.

[0027] Generally, semiconductor designs often have high rise time accuracy but low fall time accuracy. Therefore, by using only the rise time of the signal, as described above, the fall time error can be eliminated from the spatial waveform. Depending on the characteristics of the signal, it is desirable to use the transition from HIGH to LOW or LOW to HIGH, whichever has higher accuracy, as the rise time.

[0028] The spatial waveform at the bottom of Figure 6 shows the result of representing the change in the pulse period of each pulse of the pulse signal described above in time series. The spatial waveform shown here is a spatial waveform plotted from the starting point (arbitrary point) of the pulses for one rotation of the rotating body 12, with the horizontal axis representing the i-th pulse (i = 1 to n (n is an integer)) and the vertical axis representing the period (time) of each pulse.

[0029] This spatial waveform shows the rotational irregularities of the first rotating body 17 and the magnetization errors (pitch errors and eccentricity errors) of the magnetic encoder 57 as increases and decreases over time. By correcting the pulse signal using this spatial waveform, a more accurate rotational speed of the rotating body 12 can be obtained with each error minimized.

[0030] Figure 7 is an explanatory diagram illustrating an example of another definition of a pulse in a pulsed signal. The above definition of a pulse in a pulsed signal is just one example, and each repeatedly occurring HIGH voltage is defined as a pulse (T ａ１ , T ａ２ , ...) may also be defined as pulse (T ｂ１ , T ｂ２ It may also be defined as , ...). In that case, the period of one pulse can be shortened, and finer control becomes possible.

[0031] Figure 8 shows an arbitrary pulse PL from the pulse signal. ｉ The period (where i is an integer) and its pulse PL ｉ This is an explanatory diagram that schematically shows the pulses for one rotation of the rotating body 12 centered on [the object]. In the following explanation, the pulse PL shown in Figure 8... ｉ The period of a pair of HIGH voltage and LOW voltage is defined as the pulse period Tp. ｉ Let T be the period of one rotation of the hub axis 131 centered on [the specified point].

[0032] (Generation of pulse calculation values) Based on the spatial waveform described above, a period ratio waveform K is generated as a pulse calculation value, which is data for detecting changes in the rotation state. This waveform K represents the change in the pulse period Tp of each pulse relative to the rotation period T. This period ratio waveform K is generated by the pulse calculation unit 49 shown in Figure 3.

[0033] FIG. 9 is an explanatory diagram showing an example of the period ratio waveform K. The period ratio waveform K is a waveform showing the ratio Tp / T of the pulse period Tp of each pulse of the pulse signal for one rotation of the first rotating body 17 and the rotation period T for one rotation of the rotating body 12 centered on that pulse, with respect to the transition of the order of pulse generation of the pulse signal. This period ratio waveform K is a waveform for five rotations of the magnetic encoder 57, and shows a state in which the waveforms in the range from 0 to n of the number of poles of the magnetic poles of the magnetic encoder 57 (FIG. 4) are repeated five times.

[0034] (Generation of reference waveform) FIG. 10 is an explanatory diagram schematically showing the processing content of extracting the fluctuations other than the steady component that constantly occur for each rotation angle of the rotating body 12 from the period ratio waveform K and generating a rotational fluctuation waveform. The above-described period ratio waveform K includes the fluctuation P1 caused by the rolling bearings 23 and 25 shown in FIG. 1, the fluctuation P2 caused by the magnetization error of the magnetic encoder 57 and the mounting accuracy of the magnetic encoder 57, and the fluctuation P3 caused by other noises. Further, the fluctuation P1 can be classified into a fluctuation component P1 that occurs regardless of the rotation phase ａ and a steady component P1 that constantly occurs according to the rotation phase. ｂ Note that the period ratio waveform K shown here is a waveform when rotational fluctuations different from normal occur due to problems such as misalignment of the rotating body 12. The period ratio waveform K in the normal state is a waveform including only the fluctuations P2 and P3.

[0035] The values of the period ratio waveform K including the above-described fluctuations P1 (P1 ａ , P1 ｂ ), P2, and P3 are integrated for each rotation angle of the rotating body 12. This integration process is carried out by rotating the rotating body 12 a plurality of times. Then, the obtained integrated values for each rotation angle are divided by the number of integration times of the performed integration process, respectively, to obtain an average pulse calculation value averaged for each rotation angle. When this average pulse calculation value is represented by the distribution for one rotation of the rotating body 12, it becomes the composite average waveform (reference waveform) C shown in FIG. 10. This composite average waveform C includes the average waveform S1 for one rotation corresponding to the fluctuation component P1 ａ of the period ratio waveform K and the steady component P1 ｂIt can be said that this waveform is a composite of the average waveform S2 for one rotation corresponding to fluctuation P2, the average waveform S3 for one rotation corresponding to fluctuation P3, and the average waveform S4 for one rotation corresponding to fluctuation P3.

[0036] The fluctuation component P1a has periodicity, but its amplitude changes not only in width but also in both positive and negative directions, so the average waveform S1 is a flat waveform regardless of the rotation angle. Also, the fluctuation component P3 due to noise does not occur steadily at a specific rotation angle but occurs at random rotation angles, so the average waveform S4 is a flat waveform regardless of the rotation angle. On the other hand, the steady component P1 ｂ Since the fluctuation P2 occurs steadily at a specific rotation angle, it results in a waveform with a distribution corresponding to the rotation angle. The waveform obtained by combining these waveforms for each rotation angle is the composite average waveform C, which will be used as a reference waveform, as will be described in detail later. This composite average waveform C is generated by the reference generation unit 51 shown in Figure 3.

[0037] (Generation of rotational fluctuation waveform) Figure 11 is an explanatory diagram showing how the detected period ratio waveform is corrected using the composite average waveform C. Once the composite average waveform C described above is obtained, a rotational fluctuation waveform can be generated in which steady rotational fluctuations unrelated to the state change to be detected are selectively removed by applying a predetermined process using the composite average waveform C to the period ratio waveform K detected by the sensor 43.

[0038] If the aforementioned period ratio waveform K is subjected to frequency analysis, the background intensity increases due to the aforementioned fluctuation P3, and the steady-state component P1 ｂ The analysis results show that the peak intensity of the rotational Nth order (where N is an integer) increases due to the fluctuation P2. In this analysis, the steady-state fluctuation (steady-state component P1 ｂ , the residual intensity due to fluctuation P2) is strong, and the desired change in rotational state (P1) ａ ) becomes difficult to detect.

[0039] On the other hand, by applying processing using the composite average waveform C to the period ratio waveform K, a rotational fluctuation waveform with unwanted noise removed can be obtained. This rotational fluctuation waveform contains information about the change in rotational state that we want to detect, and the change in rotational state can be easily detected from this waveform or from the results of frequency analysis of the waveform.

[0040] The specific procedure for generating the rotational fluctuation waveform described above is as follows: First, the period ratio waveform K is determined for each rotation angle from the pulse signal continuously output from the sensor 43. For example, the value of the period ratio waveform K is determined for each of the rotation positions detected by the sensor, such as 0°, 10°, and 20°. As the rotating body 12 rotates multiple times, the values ​​of the period ratio waveform K at each rotation position are accumulated, and finally, each accumulated value is averaged by dividing it by the number of rotations (number of accumulations). In this way, a composite average waveform C is obtained by averaging the period ratio waveform K for each rotation angle. Next, the difference waveform (= K - C: rotational fluctuation waveform) is obtained by subtracting the value of the composite average waveform C from the average value of the period ratio waveform K obtained for each rotation angle, and representing this value as a distribution for one rotation with the rotation angle on the horizontal axis.

[0041] Frequency analysis of the obtained difference waveform reveals a steady-state fluctuation (steady-state component P1 ｂ The fluctuation P2) does not affect the analysis results, and the fluctuation P1 that we particularly want to detect ａ However, this appears in the analysis results as a peak with a high signal-to-noise ratio. By determining, for example, whether or not the peak waveform obtained from this frequency analysis is above a predetermined threshold, the change in rotational state that we want to detect can be accurately extracted. The above processing is performed by the rotational fluctuation waveform generation unit 53 shown in Figure 3, and the detection of the change in rotational state is performed by the state change detection unit 55.

[0042] Figure 12 is a control block diagram showing the process from obtaining a difference waveform from a detected pulse signal to obtaining a rotational fluctuation waveform. In this process, a period ratio waveform K (first detection waveform) is obtained based on the output signal from the sensor 43, and the obtained period ratio waveform K (first detection waveform) is integrated for each rotation angle to obtain a composite average waveform C, which is stored in the memory as a reference waveform. Then, the difference waveform is obtained by subtracting the value of the reference waveform stored in the memory from the above period ratio waveform K (first detection waveform) for each rotation angle. Alternatively, a period ratio waveform K (second detection waveform) is obtained from a new output signal output from the sensor 43, and the difference waveform is obtained by subtracting the value of the reference waveform stored in the memory from this new period ratio waveform K (second detection waveform) for each rotation angle. The difference waveform (rotational fluctuation waveform) obtained in this way is a waveform from which unnecessary fluctuations have been removed, and frequency analysis yields a frequency characteristic from which unnecessary peaks have been removed.

[0043] In this control, a reference waveform obtained from the period ratio waveform K (first detection waveform) based on the output signal from sensor 43 is used to subtract the period ratio waveform K (first detection waveform) itself, or the period ratio waveform K (second detection waveform) obtained from the subsequent output signal. In other words, a reference waveform is generated from the period ratio waveform K, which includes fluctuations caused by malfunctions, and the above period ratio waveform K is processed using the obtained reference waveform, thereby removing unnecessary fluctuations from the period ratio waveform K in real time. For example, when processing measurement data using a reference waveform, if a reference waveform based on past measurement data or calculations is prepared in advance and processing is done using this prepared reference waveform, the reference waveform may not necessarily match the timing at which the measurement data was actually acquired. Therefore, there is a risk of errors in the processing results. On the other hand, in this control, the reference waveform used for subtracting the period ratio waveform K is generated from the pulse signal from which the period ratio waveform K was obtained, or from the pulse signal output continuously with that pulse signal. In other words, the pulse signal used to generate the reference waveform, or the rotation of the rotating body while that pulse signal is being output, continues, and the reference waveform is determined using the pulse signal output continuously following the aforementioned pulse signal. Therefore, since the measurement data being evaluated and the reference waveform are information that can be considered to occur simultaneously or nearly simultaneously in time series, there is almost no change in the rotation conditions, and the above-mentioned discrepancies do not occur. Furthermore, there is no need to prepare a waveform in an ideal state with no fluctuations as a reference.

[0044] Here, the period ratio waveform K is used as an example for the first and second detection waveforms, but the system is not limited to this, and waveforms obtained by further processing the period ratio waveform K may also be used.

[0045] The following describes a specific example of control for detecting abnormalities in rotational fluctuations. Figure 13 is an explanatory diagram showing time chart 1 from the period ratio waveform to the detection of abnormalities in rotational fluctuations. In the time chart of Figure 13, the horizontal axis is elapsed time, and the vertical axis shows the processing content for each elapsed time. In this control, abnormality detection target sections SC1, SC2, SC3, SC4, ... are set intermittently and sequentially. First, the elapsed time t ０ ~t １ The output signal from the sensor is acquired during the following elapsed period t. １ ~t ２ During this process, a period ratio waveform K is generated, a reference waveform is generated from this period ratio waveform K, the reference waveform is subtracted from the period ratio waveform K to obtain a difference waveform (rotational fluctuation waveform), and this rotational fluctuation waveform is frequency-analyzed using FFT or the like to determine if it is abnormal. ０ ~t ２ During the period up to (the period enclosed by the dashed line TM), the elapsed time t ０ ~t １ The abnormality detection determination for section SC1 is completed. The same process is then repeated sequentially for SC2, SC3, SC4, and so on.

[0046] Figure 14 is an explanatory diagram showing a time chart 2 from the period ratio waveform to the determination of rotational fluctuation abnormalities. The time chart 2 in Figure 14 is an example of control that can be executed even when the computational processing power of the control unit 15 is relatively low. In this control, first, the elapsed time t ０ ~t １ The output signal from the sensor is acquired during the following elapsed period t. １ ~t ２ During this time, a period ratio waveform K is generated, and a reference waveform is generated from this period ratio waveform K. Next, the elapsed period t ２ ~t ３ During the period t, the output signal from the sensor is acquired to generate a period ratio waveform K, while the elapsed period t １ ~t ２ The reference waveform obtained during this time is subtracted from the period ratio waveform K to obtain the difference waveform (rotational fluctuation waveform). Then, the elapsed time t ３ ~t ４The rotational fluctuation waveform is frequency-analyzed using FFT (Fast Fourier Transform) or similar methods to detect anomalies. ０ ~t ４ During the period up to, the elapsed time t ２ ~t ３ The abnormality determination of the section SC1 targeted for abnormality detection is completed. In this case, the abnormality determination of SC1 is completed after elapsed time t ０ ~t １ A reference waveform based on information from that period will be used, but since it is close to the SC1 period, no significant changes will occur, and no non-conformities will occur.

[0047] Furthermore, elapsed time t ４ ~t ５ During the period t, the output signal from the sensor is acquired to generate a period ratio waveform K, while the elapsed period t １ ~t ２ The reference waveform obtained during this time is subtracted from the period ratio waveform K to obtain the difference waveform (rotational fluctuation waveform). Then, the elapsed time t ５ ~t ６ The rotational fluctuation waveform is analyzed by frequency analysis using FFT, etc., to determine if an anomaly occurs. ４ ~t ６ During the period up to, the elapsed time t ４ ~t ５ The abnormality determination of the section SC2 targeted for abnormality detection is completed. Similarly, t ６ ~t ８ During the period up to, the elapsed time t ６ ~t ７ The abnormality detection determination for the section SC3 in the above section is completed. The same process is then repeated sequentially. ０ ~t ８ During the period up to (the period enclosed by the dashed line TM), the elapsed period t １ ~t ２ The abnormality detection for the target sections SC1, SC2, and SC3 is completed using the reference waveform obtained between them.

[0048] In this control, abnormality detection for the periods of SC1, SC2, and SC3 is performed over the elapsed period t. １ ~t ２It is carried out based on the result of commonly using the reference waveform obtained among them. Thereby, it is possible to perform abnormality determination while reducing the calculation load of the control unit.

[0049] FIG. 15 is an explanatory diagram showing a time chart 3 until abnormality determination of rotational fluctuation is carried out from a period ratio waveform. The time chart 3 in FIG. 15 is a control example that can be executed even when the calculation processing ability of the control unit 15 is lower. In this control, first, the output signal from the sensor is acquired during the elapsed time t ０ to t １ . Next, during the elapsed period t １ to t ２ , the period ratio waveform K is generated, and a reference waveform is generated from this period ratio waveform K. Then, the output signal from the sensor is acquired during the elapsed time t ２ to t ３ , the period ratio waveform K of the output signal is generated during the elapsed period t ３ to t ４ , and the reference waveform generated during t １ to t ２ is subtracted from the period ratio waveform K to obtain a difference waveform (rotational fluctuation waveform). Also, the difference waveform during this period from t １ to t ２ is subjected to frequency analysis.

[0050] Next, the output signal from the sensor is acquired during the elapsed period t ４ to t ５ , the period ratio waveform K of the output signal is generated during t ５ to t ６ , and the reference waveform generated during t １ to t ２ is subtracted from the period ratio waveform K to obtain a difference waveform (rotational fluctuation waveform). Also, the difference waveform during this period from t ４ to t ５ is subjected to frequency analysis.

[0051] Similarly, the output signal from the sensor is acquired during the elapsed period t ６ to t ７ , the period ratio waveform K of the output signal is generated during t ６ to t ７ , and the reference waveform generated during t １ to t ２Subtract the reference waveform generated during the period to obtain a difference waveform (rotation fluctuation waveform). Also, perform frequency analysis on the difference waveform during this period from t ６ to t ７ The result of frequency analysis of the difference waveform during the period from elapsed time t ８ to t ９ is stored. During the elapsed period from t ３ to t ４ (corresponding to SC1), t ５ to t ６ (corresponding to SC2), and t ７ to t ８ (corresponding to SC3), average the results of the frequency analysis obtained, and perform abnormality determination for the periods of SC1, SC2, and SC3.

[0052] In this control, the abnormality determination for the periods of SC1, SC2, and SC3 is performed based on the result of commonly using the reference waveform obtained during the elapsed period from t １ to t ２ . Also, by performing frequency analysis separately for each period, abnormality determination can be performed while further reducing the calculation burden of the control unit. [[ID=,29]]

[0053] In each of the above control examples, by setting the subtraction process of the reference waveform and the section of frequency analysis, which have a large calculation burden, outside the abnormality detection target section, efficient calculation can be performed even when the calculation ability of the control unit 15 is relatively low. Also, since the difference waveform (rotation fluctuation waveform) is obtained using the reference waveform generated based on the information under the same conditions as when measuring the period ratio waveform K, or the reference waveform generated under conditions close to those during measurement, the accuracy of the reference waveform is high and more accurate abnormality determination is possible.

[0054] (Case of a rotational state including fluctuations caused by internal bearing defects) Next, a control example in the case where the detection signal from the sensor includes fluctuations caused by internal bearing defects in addition to the fluctuations such as misalignment described above will be described.

[0055] FIG. 16 is an explanatory diagram showing the period ratio waveform K in the case of a rotational state having misalignment and internal bearing defects. In the period ratio waveform K, in addition to the fluctuations P1 (P1 ａ , steady component P1 ｂ ), P2, and P3, there are also fluctuations P ｉｎThese can sometimes be superimposed. Internal bearing defects can occur at frequencies deviating from the rotational order, and in such cases, frequency analysis of the period ratio waveform K will show peaks at frequencies other than the rotational order. For example, if damage such as delamination or indentation occurs on the bearing raceway surface, the rotational speed will fluctuate in conjunction with the damage. In the frequency characteristics of this, peaks corresponding to the spatial frequency of the defect (how many times the defect impacts per rotation), calculated using various variables including the diameter d (mm) of the bearing rolling elements, the pitch circle diameter D (mm) of the rolling elements, the number of rolling elements Z, and the contact angle α (rad), appear regardless of the rotational order frequency.

[0056] Figure 17 shows the variation P caused by internal bearing defects. ｉｎ This is an explanatory diagram showing the average period ratio waveform. The fluctuation P caused by the bearing internal defect described above. ｉｎ The period ratio waveform, if its spatial frequency is an integer, maintains approximately the original waveform shape when averaged over each rotation. On the other hand, when the spatial frequency deviates from an integer, the waveform averaged over each rotation is smoothed out, resulting in a flat waveform shape. Therefore, as shown in Figure 10, the fluctuation P ｉｎ When the period ratio average waveform is averaged for each rotation angle, it results in a flat frequency response and no longer affects the reference waveform. Therefore, even if the reference is subtracted from the period ratio waveform, the result of frequency analysis of the difference waveform will not show fluctuation P. ｉｎ This will result in a peak occurring.

[0057] Figure 18 is a schematic diagram illustrating the frequency characteristics of the period ratio waveform K in the case of misalignment and internal bearing defects. The frequency characteristics illustrated here include the peak (variation P) of the rotational order frequency. １ａ , P ２ , steady portion P １ｂ In addition to the rotation order (integer), there are also peaks at frequencies other than the rotation order (variation P). ｉｎ ) is occurring. From this frequency characteristic, the desired variation (P １ａ , P ｉｎ To selectively extract the characteristics of each rotation order, one could consider using a bandpass filter that selectively attenuates the steady-state components of each rotation order.

[0058] Figure 19 is an explanatory diagram showing the results of frequency analysis of the period ratio waveform K shown in Figure 16 after processing it with bandpass filters BPF1 and BPF2. Bandpass filter BPF1, shown on the left side of Figure 19, is a filter that selectively attenuates the spectrum of the rotation order frequency (attenuation: low). Bandpass filter BPF2, shown on the right side, has a greater attenuation of the rotation order frequency compared to BPF1 (attenuation: high). Filtering the period ratio waveform K using these bandpass filters BPF1 and BPF2 can attenuate the intensity for the rotation order and frequencies close to it, but it is difficult to set the frequency filter constants (characteristics) so that the stationary component that does not need to be detected becomes zero.

[0059] On the other hand, as shown in Figure 11, when the difference waveform (rotational fluctuation waveform) is obtained by subtracting the reference averaged for each rotation angle from the period ratio waveform K, the peak heights of the rotation order and frequencies close to the rotation order in the waveform after frequency analysis of the difference waveform maintain a sufficient signal-to-noise ratio.

[0060] Figure 20 is an explanatory diagram showing the results of determining the frequency characteristics from the difference waveform in a rotational state with misalignment and internal bearing defects. Figure 20 shows the result of frequency analysis of the difference waveform (K-C) obtained by subtracting the previously determined composite average waveform C from the period ratio waveform K when there are peaks at the rotational order and a frequency shifted from the rotational order. In this case, the peak of the fluctuation to be detected (P １ａ , P ｉｎ The peaks in the fluctuations are selectively extracted and clearly displayed with a high signal-to-noise ratio. In other words, by detecting the peaks of the fluctuations using this method, each peak can be reliably detected with a lower peak detection threshold SHD than the method described above. Thus, the process of subtracting the reference waveform (composite average waveform C) from the period ratio waveform K can reliably extract peaks that cannot be extracted by the bandpass filters BPF1 and BPF2 described above.

[0061] Figure 21 is an explanatory diagram showing the result of applying a low-pass filter (LPF) to the difference waveform (K-C) shown in Figure 11. The difference waveform (K-C) shown here is a waveform that includes fluctuations due to misalignment, and high-frequency fluctuations that are not related to anomaly detection may be superimposed. In that case, these high-frequency fluctuations are removed by the low-pass filter. This makes it easier for operators to identify the fluctuations they want to detect when they view and check the difference waveform (K-C).

[0062] Figure 22 is an explanatory diagram showing the result of applying a high-pass filter (HPF) to the difference waveform (K-C) obtained by subtracting the composite average waveform C from the period ratio waveform K shown in Figure 16. The difference waveform (K-C) shown here is a waveform that includes fluctuations due to misalignment and fluctuations due to internal bearing defects. In this case, if it is particularly important to detect internal bearing defects occurring on the high-frequency side, removing the fluctuations on the low-frequency side with a high-pass filter will yield a waveform with small fluctuations due to the internal bearing defects. As a result, when an operator visually checks the difference waveform (K-C), the undulations of the waveform described above are removed, making it easy to identify the cause of vibration and perform other related tasks. Alternatively, a band-pass filter that extracts or removes only the necessary frequency band may be used instead of the low-pass and high-pass filters described above.

[0063] As described above, prior to measurement, the period ratio waveform K, which is a pulse calculation value, is first integrated for each rotation angle of the rotating body 12, and the obtained integrated values ​​are averaged for each rotation angle to generate a reference waveform that represents the distribution for one rotation of the rotating body 12. This reference waveform is stored in the memory of the control unit 24. Then, the period ratio waveform K based on the pulse signal newly output from the sensor 43 is processed using the stored composite average waveform C reference waveform to obtain the rotation fluctuation waveform. In this way, by detecting changes in the rotation state of the rotating body 12 based on the rotation fluctuation waveform from which noise has been selectively removed, the accuracy of anomaly detection can be improved.

[0064] Figure 23 is a schematic diagram illustrating the difference in anomaly detection accuracy depending on the reference waveform. It compares the frequency characteristics of a waveform obtained by subtracting a pre-prepared reference waveform from a period ratio waveform K having the frequency characteristics shown in Figure 18, with the frequency characteristics of a waveform obtained by subtracting a reference waveform obtained from the sensor output simultaneously with or immediately after the sensor output data. The former result shows that unnecessary peaks remain for anomaly detection, requiring a relatively high threshold SHD for detection. On the other hand, the latter result, processed with a more accurate reference waveform, reliably removes unnecessary peaks for anomaly detection, and obtains the peaks necessary for anomaly detection with a high S / N ratio. Therefore, even if the threshold SHD used for anomaly detection is set low, the desired peaks can be stably detected.

[0065] The anomaly detection described above is not limited to determination from frequency characteristics; for example, anomalies may also be detected by visually inspecting the profile obtained by subtracting a reference waveform from a rotational fluctuation waveform such as the period ratio waveform K, or by setting appropriate judgment conditions. The profile obtained by subtracting the reference waveform is easier to evaluate than the period ratio waveform K because the waveform characteristics are more clearly expressed in the profile obtained by subtracting the reference waveform.

[0066] The state change detection system 100 described above can be used to detect the rotational state of the rotating body 12 and monitor various state changes. Specific examples of state changes include, for example, peeling or indentation on the raceway surfaces and rolling element surfaces of the rolling bearings 23 and 25, loosening of the fastener 21, opening of the flange portions 29 and 33, and loss of preload in the rolling bearings 23 and 25. The following describes in more detail an example of applying the state change detection system 100 to the above applications.

[0067] <Hub Unit Bearing Damage Detection System> (First Configuration Example) Figure 24 is a schematic diagram of the first configuration example of the hub unit bearing bearing damage detection system 300. In this specification, "axial inner" with respect to the hub unit bearing refers to the vehicle body side of the hub unit bearing when it is mounted on the vehicle body, and means the right side indicated by the arrow Ax_in in Figure 24. Also, "axial outer" refers to the wheel side of the hub unit bearing when it is mounted on the vehicle body, and means the left side indicated by the arrow Ax_out in Figure 24.Therefore, the bearing portion, outer ring raceway and inner ring raceway located on the inside are also referred to as the inner row bearing portion, inner row outer ring raceway and inner row inner ring raceway, and the bearing portion, outer ring raceway and inner ring raceway located on the outside are also referred to as the outer row bearing portion, outer row outer ring raceway and inner row inner ring raceway.

[0068] The hub unit bearing 111 shown in Figure 24 is a hub unit bearing for a driven wheel and comprises a fixed-side member, an outer ring (hub outer ring) 113, a rotating-side member, a hub 115 (first rotating body, hub inner ring), a plurality of rolling elements 117, and a rotation detection unit 119. In Figure 24, the hub 115 of the hub unit bearing 111 and the sensor 121, which will be described later, are shown in a horizontal cross-section. The bearing damage detection system 300 for the hub unit bearing 111 includes the hub unit bearing 111, a pulse signal generation unit 123 that outputs the detection signal from the sensor 121 of the rotation detection unit 119 of the hub unit bearing 111 as a pulse signal, and a control unit 124. Although a hub unit bearing for a driven wheel is shown here, bearing damage can be detected in a similar manner for a hub unit bearing for a drive wheel.

[0069] The outer ring 113 has a stationary flange 125 on its outer circumference and outer ring raceways 127 and inner ring raceways 129 on its inner circumference. When in use, the outer ring 113 is prevented from rotating while supported by the suspension device by connecting and fixing the stationary flange 125 to a knuckle of the suspension device (not shown).

[0070] The hub 115 is composed of a hub shaft 131 and an inner ring 133 that fits onto the hub shaft 131 and is fixed by crimping, and is arranged coaxially (concentrically) with the outer ring 113 on the radially inner side of the outer ring 113.

[0071] The hub shaft 131 is provided with a ring-shaped mounting flange 135 that extends radially outward from the axially outward protruding portion of the outer ring 113, for fixing a wheel (driven wheel) and a braking rotating member (second rotating body) such as a disc rotor (not shown). The braking rotating member has a recess (not shown) corresponding to the recess 39 (Figure 1) for the aforementioned spigot fitting, and the hub shaft 131 has a protruding shaft 38 that is inserted into the recess.

[0072] The mounting flange 135 is provided with multiple through holes 135a, and hub bolts 137 (fasteners) are serrated into each through hole 135a. Alternatively, the multiple through holes 135a can be made into female threaded holes, and the hub bolts can be screwed into them to fix the mounting flange 135 to a wheel and a braking rotating member such as a disc rotor.

[0073] On the outer circumferential surface of the hub shaft 131, the portion facing the outer ring raceway 127 of the outer row is provided with the inner ring raceway 139 of the outer row. Furthermore, on the outer circumferential surface of the hub shaft 131, the axial inner end portion facing the outer ring raceway 129 of the inner row is provided with a small-diameter stepped portion 141. The hub shaft 131 partially constitutes the outer circumferential surface of the small-diameter stepped portion 141 and has a crimping portion 143 at its axial inner end portion that deforms radially outward to fix the inner ring 133 by crimping.

[0074] The inner ring 133 has an inner ring raceway 145 on its outer circumferential surface in the portion facing the outer ring raceway 129 of the inner row. The inner ring 133 is fitted onto the small-diameter stepped portion 141 of the hub shaft 131 with its axial outer end face abutting against the stepped surface of the small-diameter stepped portion 141, and is crimped and fixed to the hub shaft 131 by a crimping portion 143, which is a crimping portion formed by the axial inner end of the small-diameter stepped portion 141 being deformed radially outward.

[0075] The rolling elements 117 are provided to roll freely between the outer ring raceway 127 and the inner ring raceway 139 of the outer row, and between the outer ring raceway 129 and the inner ring raceway 145 of the inner row, while being held by their respective retainers 147.

[0076] Furthermore, the outer ring raceway 127 of the outer row, the inner ring raceway 139 of the outer row, and the rolling elements 117 form the bearing portion 149A of the outer row, and the inner ring raceway 129 of the inner row, the inner ring raceway 145 of the inner row, and the rolling elements 117 form the bearing portion 149B of the inner row.

[0077] A seal ring 151 is fixed to the axially outer end of the inner circumferential surface of the outer ring 113. The seal ring 151 closes the axially outer end opening of the internal space 153, which is located between the inner circumferential surface of the outer ring 113 and the outer circumferential surface of the hub shaft 131 and contains a plurality of rolling elements 117. The seal ring 151 slides against the large-diameter stepped portion of the outer circumferential surface of the hub shaft 131, which is axially outward from the inner ring raceway 139 of the outer row.

[0078] The rotation detection unit 119 is an axial type sensor disposed near the inner row bearing portion 149B, that is, at the axial inner end of the hub unit bearing 111, to detect the rotational speed of the hub 115, and comprises a magnetic encoder 155 and a sensor 121.

[0079] The magnetic encoder 155 consists of a support ring 155a and an encoder body 155b. The support ring 155a is formed in an L-shape in cross-section and an annular shape overall by press-forming a magnetic metal plate such as a ferritic stainless steel plate such as SUS430 or a rolled steel plate such as SPCC. The axial outer portion of the support ring 155a is fitted onto and fixed to the inner ring 133.

[0080] The encoder body 155b is made entirely of a ring-shaped permanent magnet formed by mixing a magnetic material such as ferrite powder into rubber or thermoplastic resin, and is attached and fixed to the inner side surface of the ring portion of the support ring 155a, which is bent radially inward. The inner side surface of the encoder body 155b is magnetized with alternating S poles and N poles at equal pitches in the circumferential direction.

[0081] Sensor 121 is a magnetic sensor that detects the rotation of the hub 115, particularly the rotation of the inner portion of the hub 115. Sensor 121 is positioned with its detection surface 121a facing the magnetic encoder 155 and is fixed to a side cover 159 that closes the inner opening of the outer ring 113. Sensor 121 detects the rotation of the hub 115 by detecting the change in magnetism in the detection area DR of the magnetic encoder 155 facing the detection surface 121a as the hub 115 rotates. In other words, sensor 121 and the magnetic encoder 155 function as a rotation sensor, and this rotation sensor and the pulse signal generation unit 123 constitute the rotation detection unit 119.

[0082] The rotation sensor is not limited to the above configuration; it may also be a combination of a cylindrical encoder and a radial sensor. Furthermore, it is not limited to a magnetic encoder; other types such as optical encoders and proximity sensors may be used to detect rotation, or a gear-shaped encoder or a gear-shaped exciter ring used in trucks may be used.

[0083] Sensor 121 may be, for example, an active wheel speed sensor provided on the hub unit bearing 111. In that case, a detection element such as a Hall IC element or MR element whose electrical properties change in response to magnetism can be used as sensor 121. The active wheel speed sensor faces a magnetic encoder and outputs a HIGH voltage when the magnetic flux density is below a threshold, and outputs a LOW voltage when the poles of the magnetic encoder approach and the magnetic flux density exceeds the threshold. The active wheel speed sensor generates a pulse wave corresponding to the rotation speed of the tire. Generally, this pulse wave is sent to an on-board controller and used for ABS (Anti-lock Braking System) and traction control, but by using this pulse wave for damage detection, the need to add a separate rotation sensor is eliminated.

[0084] The sensor 121 outputs a rotation detection signal to the pulse signal generation unit 123. The pulse signal generation unit 123 generates a pulse signal based on the input detection signal and outputs the generated pulse signal to the control unit 124.

[0085] Figure 25 is a functional block diagram of the control unit 124. The control unit 124 comprises a rotational fluctuation extraction unit 124A, a frequency analysis unit 124B, a peak intensity calculation unit 124C, and a damage detection unit 124D. The rotational fluctuation extraction unit 124A has the same function as the rotational fluctuation extraction unit 47 shown in Figure 3 above. The frequency analysis unit 124B, the peak intensity calculation unit 124C, and the damage detection unit 124D function as a state change detection unit 55. The rotational fluctuation extraction unit 124A extracts the rotational fluctuation of the rolling bearing from the rotation signal output from the rotation sensor and generates a rotational fluctuation signal. The frequency analysis unit 124B performs frequency analysis on the waveform of the rotational fluctuation signal to determine the frequency characteristics. The peak intensity calculation unit 124C determines the peak intensity corresponding to bearing damage from the frequency characteristics. The damage detection unit 124D detects bearing damage of the rolling bearing based on the peak intensity.

[0086] The control unit 124 executes a procedure for determining damage to the hub unit bearing 111 based on the pulse signal input from the pulse signal generation unit 123. This control unit 124 is configured as a computer equipped with the aforementioned processor, memory, storage, and other storage devices. In this case, the functions of the rotation detection unit 119 shown in Figure 24, the pulse signal generation unit 123, and each unit shown in Figure 25 can be realized by the processor executing a predetermined program stored in the storage device. Furthermore, the control unit 124 is not limited to being directly or wirelessly connected to the hub unit bearing 111 and the rotation detection unit 119, but may also be connected via communication such as a network. In that case, the determination of damage to the hub unit bearing 111 can be performed from a remote location, improving the convenience of management. It also becomes easier to manage multiple hub unit bearings 111 collectively.

[0087] In the bearing damage detection system 300 for the hub unit bearing 111 of this embodiment, the control unit 124 detects damage occurring in the outer ring raceway or inner ring raceway with high accuracy based on the pulse signal generated by the rotation detection unit 119. This pulse signal generally includes changes in frequency depending on the rotation speed of the tire, as well as rotational irregularities of the tire, magnetization errors due to the magnetization pitch and magnetization eccentricity of the magnetic encoder, and deformation when the magnetic encoder is pressed into the inner ring. Furthermore, if the bearing raceway surface is damaged, the pulse signal also includes errors due to the relative displacement of the outer and inner rings caused by the damage, and the relative displacement due to the impact when the rolling elements enter and exit the damaged area. When all of this various information is included in the pulse signal, it becomes difficult to extract only the damage information. However, in this bearing damage detection system 300, by removing the above-mentioned noises included in the pulse signal, damage information can be extracted with high accuracy, enabling accurate evaluation.

[0088] Next, we will explain the process of generating the pulse signal described above based on the detection signal from sensor 121. With respect to sensor 121, a pulse signal is generated in the same way as shown in Figures 4 and 5 above.

[0089] Figure 26 is a schematic diagram illustrating the occurrence of vibration when delamination occurs in a part of the outer ring raceway 127. For example, consider a situation where a radial load Pr is applied from the inner ring 133 to the outer ring 113 in the vertically upward direction, and a defective region Ad exists on the vertically upward side of the outer ring raceway 127 where delamination has occurred. In this case, when the rolling element 117 that rolls between the inner ring 133 and the outer ring 113 enters the defective region Ad, the internal gap between the outer ring raceway 127 and the inner ring raceway 145 widens due to the delamination, causing the rolling element 117 moving across the delamination surface to float. As the floating rolling element 117 moves along the rotational direction Ro of the inner ring 133 through the defective region Ad, reaches the edge of the defective region Ad, and re-enters the space between the outer ring raceway 127 and the inner ring raceway 145, which are free from separation, the rolling element 117 collides with the inner ring 133 and the outer ring 113, generating an impact load that pushes the inner ring 133 downwards. Note that the load on the inner ring occurs both horizontally and downwards, but here we will focus on the downward load. Repeated collisions of the rolling element 117 generate vibrations in the vertical (and horizontal) directions. In other words, as the inner ring 133 rotates, the inner ring 133 and the outer ring 113 move relative to each other vertically at the defect cycle, and damage can be detected by detecting this relative vertical displacement with the sensor 121.

[0090] Figure 27 is an explanatory diagram showing the change in sensor output due to the relative displacement between the inner ring 133 and the outer ring 113 caused by the separation of the outer ring raceway. In Figure 27, the case where separation occurs in the outer ring raceway 127 of the outer row bearing section 149A, which has an outer ring raceway 127 and an inner ring raceway 139, and the inner row bearing section 149B, which has an outer ring raceway 129 and an inner ring raceway 145. Note that this hub unit bearing is shown as a hub unit bearing for a drive wheel.

[0091] Here, we will explain using the example of applying a radial load Pr (preload) directed vertically upward from the hub shaft 131, which is the inner ring of the bearing section 149A, to the outer ring 113. However, damage can be detected using the same procedure even when no preload is applied. When the rolling element 117A is floating in the defective area Ad where delamination has occurred in the outer ring raceway 127, the relative vertical velocity VR between the hub shaft 131 and the outer ring 113 is measured in the detection area DR on one horizontal side where the sensor (not shown) is located, and on the other horizontal side opposite to it. ０ And the relative vertical velocity VL on the other side ０ This is equivalent to:

[0092] Next, when the rolling element 117A reaches the edge of the defective region Ad and rides up between the outer ring raceway 127 and the inner ring raceway 139, which are free from separation, an impact load Pinp is applied to the hub shaft 131. Then, the relative velocity between the hub shaft 131 and the outer ring 113 in the detection region DR is the upward relative velocity VR in the floating state. ０ Furthermore, a reverse (downward) velocity Vp is added due to the displacement caused by the impact load Pinp, resulting in a relative velocity VR in the floating state. ０ Lower speed VR １ This is the result. On the other hand, on the other side of the detection region DR in the horizontal direction, the relative velocity VL ０ A velocity Vp is added due to an impact load Pinp in the same direction, resulting in a relative velocity VL in the floating state. ０ Larger speed VL １ This is the result.

[0093] Then, as the hub shaft 131 rotates further and the rolling elements 117A transition back to their original rolling state between the outer ring raceway 127 and the inner ring raceway 139, a return load Pbk acts on the hub shaft 131. Then, the relative velocity between the hub shaft 131 and the outer ring 113 in the detection region DR is equal to the upward relative velocity VR in the floating state. ０ Then, a velocity Vq in the same direction (upward) is added due to the displacement caused by the return load Pbk, resulting in a relative velocity VR in the floating state. ０ Higher speed VR ２ This is the result. On the other hand, on the other side of the detection region DR in the horizontal direction, the relative velocity VL ０ Then, a return load Pbk in the opposite direction (upward) adds a velocity Vq, resulting in a relative velocity VL in the floating state.０ Smaller speed VL ２ This is the result.

[0094] Figure 28 is a schematic diagram illustrating the changes in the sensor output signal from sensor 121 when the rolling element transitions from a floating state to a riding state and back to its original rolling state. The speed detected in the detection region DR is VR ０ , VR １ , VR ２ It changes in the following order. This VR ０ From VR １ ,VR ２ By capturing changes in the sensor output signal due to changes in velocity, the presence of damage can be identified.

[0095] (Procedure for Damage Determination) Figure 29 is a flowchart showing the procedure for determining damage to the hub unit bearing 111. Each of the following steps is performed based on commands from the control unit 124 shown in Figures 24 and 25. First, the rotation of the hub shaft 131, which is driven to rotate the hub unit bearing 111, is detected by the sensor 121, and the sensor output signal, which is a rotation signal output from the sensor 121, is acquired (S11). This sensor output signal is converted into a pulse signal by the pulse signal generation unit 123.

[0096] Then, the changes in the period of each pulse (for example, T1 to T7) of the pulse signal shown in Figure 6 above are converted into a spatial waveform represented in this order in a time series (S12). In this configuration, when the pole of the magnetic encoder 155 reaches the sensor 121, the sensor output signal changes from a HIGH voltage to a LOW voltage, so the timing of the transition from HIGH voltage to LOW voltage is used as the rising edge timing described above.

[0097] Based on the spatial waveform described above, a pulse period waveform WF1 is generated as data for damage detection, representing the transition of the pulse period Tp of each individual pulse (S13). In addition, for each pulse of the pulse signal, a rotation period waveform WF2 is generated, representing the transition of the period T of one rotation centered on that pulse (S14).

[0098] Figure 30 is an explanatory diagram showing an example of a waveform from the pulse output signal to the generation of a rotational fluctuation signal by extracting the rotational fluctuation of the hub shaft 131. Note that the waveform data described below is for the purpose of explaining the content of the signal processing and is not necessarily information obtained from an actual automobile hub unit bearing. The horizontal axis of each waveform is, as an example, a spatial value corresponding to the number of pulses for 5 rotations of the hub shaft 131. After generating the pulse period waveform WF1 and rotation period waveform WF2 described above, a period ratio waveform WF3 (= WF1 / WF2) representing the ratio Tp / T of the pulse period Tp to the rotation period T is obtained (S15).

[0099] While the pulse period waveform WF1 and rotation period waveform WF2 mentioned above had time on their vertical axes, the vertical axis of the period ratio waveform WF3 represents the ratio value to one rotation. Therefore, as mentioned earlier, the effect of rotation speed error is eliminated in the period ratio waveform WF3.

[0100] Next, the period ratio waveform WF3 is smoothed to obtain the period ratio smoothed waveform WF4 (S16). Smoothing can be calculated, for example, by using a moving average of seven points before and after the original waveform. This period ratio smoothed waveform WF4 is smoothed in the same way as the period ratio waveform WF3 which has been subjected to a low-pass filter, allowing low-frequency fluctuations such as rotational irregularities to be extracted.

[0101] Generally, low-pass filters (LPFs) include IIR (infinite impulse response) and FIR (finite impulse response). However, IIR has a delay element, which may cause a phase lag in the averaged data. While FIR provides stable results, it is computationally intensive, and unless high-performance computing elements (such as expensive CPUs) are used, a phase lag may occur in the averaged data. Therefore, here, data smoothing is performed using a moving average, which requires less computation and allows for high-speed processing. However, various smoothing methods should be used as appropriate depending on the situation.

[0102] Then, the difference between the period ratio waveform WF3 and the period ratio smoothed waveform WF4 is calculated to obtain the period ratio difference waveform WF5 (= WF3 - WF4) from which the fluctuations have been extracted (S17). This period ratio difference waveform WF5 is a waveform from which low-frequency fluctuations such as rotational unevenness have been removed, but displacement due to changes in the rolling element position and magnetization errors of the magnetic encoder still remain. In addition, errors due to bearing damage when the bearing raceway surface, etc., are damaged also remain. Note that, depending on the conditions, the period ratio waveform WF3 may be used as the period ratio difference waveform WF5 without subtracting the period ratio smoothed waveform WF4 from the period ratio waveform WF3.

[0103] Furthermore, if the computer has sufficient processing power, a low-pass filter (LPF) that does not cause phase lag may be applied to the period ratio waveform WF3 to calculate the period ratio smoothed waveform WF4. Alternatively, a high-pass filter (HPF) that does not cause phase lag may be applied to the period ratio waveform WF3 to calculate the period ratio difference waveform WF5.

[0104] Next, the average value is calculated from the period ratio difference waveform WF5 for each rotation position (rotation angle), that is, for each pole of the magnetic encoder 155 (S18). In other words, although the period ratio difference waveform WF5 shown in Figure 30 represents a range of 5 rotations of the hub shaft 131, for example, in a magnetic encoder 155 with 48 poles per rotation, the values ​​of the period ratio difference waveform WF5 are extracted for each pole and averaged to obtain a period ratio average value distribution WF6 (reference waveform) that represents the distribution of average values ​​for the number of rotations for each pole.

[0105] Figure 31 is an explanatory diagram showing an example of a period ratio mean distribution WF6, which is the average value of the period ratio difference for each pole of a magnetic encoder. In the period ratio mean distribution WF6, the deviation from the 0 level on the vertical axis represents the magnetization error. In Figure 31, the line (not shown) connecting the average values ​​for each pole of the period ratio mean distribution represents the period ratio mean waveform.

[0106] Next, the period ratio fluctuation waveform WF7 (rotational fluctuation waveform), which represents the difference between the period ratio difference waveform WF5 and the period ratio average value waveform WF6 described above, is determined (S19). The period ratio fluctuation waveform WF7 is obtained by subtracting the values ​​for each pole shown in the period ratio average value waveform WF6 from the values ​​for each pole in the period ratio difference waveform WF5, for each corresponding pole. The period ratio difference waveform WF5 described above corresponds to the "detection waveform," the period ratio average value waveform WF6 corresponds to the "reference waveform," and the period ratio fluctuation waveform WF7 corresponds to the "rotational fluctuation waveform." Figure 32 is an explanatory diagram that schematically shows the period ratio fluctuation waveform WF7. In this period ratio fluctuation waveform WF7, the magnetization error of the magnetic encoder 155 is removed. In other words, since the period ratio fluctuation waveform WF7 removes fluctuations due to changes in rotational speed, rotational unevenness, and magnetization errors, the fluctuations that appear here can be said to be due to displacements associated with changes in the position of the rolling elements, and displacements associated with bearing damage when the bearing is damaged.

[0107] Furthermore, the period ratio fluctuation waveform WF7 may be a waveform obtained by converting the vertical axis from the period ratio to the velocity fluctuation through geometric calculation (S20). In that case, it becomes easier to intuitively grasp the level of fluctuation as the magnitude of the velocity. The processes from S12 to S19 and S20 described above are performed by the rotational fluctuation extraction unit 124A of the control unit 124 shown in Figure 25.

[0108] Next, the frequency analysis unit 124B performs a frequency analysis on the period ratio fluctuation waveform WF7 described above to determine the frequency characteristics (S21). Various analysis methods can be used for frequency analysis, such as the discrete Fourier transform method (FFT) or the maximum entropy method.

[0109] Figure 33 is a schematic diagram illustrating the frequency characteristics WF8 obtained by performing FFT processing on the period ratio fluctuation waveform WF7. The period ratio fluctuation waveform WF7 includes speed fluctuations associated with damage to the bearing raceway surface if such damage occurs. In the frequency characteristics WF8, if damage such as delamination or indentation occurs on the raceway surface, a peak corresponding to the spatial frequency of the defect (how many times the defect impacts per rotation) calculated by each variable including the diameter d (mm) of the rolling element, the pitch circle diameter D (mm) of the rolling element, the number of rolling elements, and the contact angle α (rad) of the bearing appears. If such a peak due to bearing damage exceeds a threshold, it is determined that the bearing has failed (S22). Note that the same frequency characteristics WF8 can be obtained even if the vertical axis of the period ratio fluctuation waveform WF7 is converted to speed fluctuation.

[0110] In the frequency response WF8 shown in Figure 33, the main peaks caused by bearing damage appear as, for example, the first-order defect peak Pk1 and the second-order defect peak Pk2. The frequencies at which the peaks appear can generally be determined by calculation depending on the nature of the defect. Numerous peaks appear in the low-frequency region (e.g., the spatial band BD0), but these are due to the system's inherent frequencies and not to bearing damage. When determining bearing damage using peaks Pk1 and Pk2 caused by bearing damage, for example, the first-order outer ring defect peak Pk1 is relatively close to the system's inherent frequency and therefore easily mixes with the peak of the system's inherent frequency. On the other hand, if the second-order defect peak Pk2 is far away and less likely to mix, the determination may be made using the second-order defect peak Pk2. In addition, around the first-order defect peak Pk1 and the second-order defect peak Pk2, sidebands may occur due to modulation with the system's low-frequency natural vibrations or other factors unrelated to the detection target. Therefore, to determine the peak intensity due to bearing damage, spatial bands BD1 and BD2 of specific lengths may be set for peaks Pk1 and Pk2, respectively, and bearing damage may be determined according to the sum of the peak intensities within each spatial band BD1 and BD2. As the contact angle of the rolling elements changes slightly depending on the preload and load conditions mentioned above, the theoretical defect frequency and the measured peak frequency may differ by several percent. Even in such cases, setting spatial bands BD1 and BD2 of specific lengths as described above can reliably prevent detection errors. In addition to the above example of determination, the extraction and determination of desired peaks and their intensities can be performed using appropriate algorithms depending on the situation.

[0111] Figure 34 is an explanatory diagram showing the frequency characteristics of a pulse period waveform WF1, which is shown as a reference example. Since the pulse period waveform WF1 contains defect information other than bearing damage, such as rotational irregularities and magnetization errors, frequency analysis of it will result in the appearance of many peaks, including rotational Nth order peaks (where N is an integer) and peaks of system-specific frequencies. In that case, compared to the case shown in Figure 33, it becomes difficult to select and extract peaks due to bearing damage defects, and a decrease in the accuracy of bearing damage determination is unavoidable. Therefore, by removing defect information other than bearing damage from the pulse period waveform WF1, as in this method, bearing damage information can be easily extracted from the results of frequency analysis, and the accuracy of damage determination can be improved. The process of calculating peak intensity from the above frequency characteristics is performed by the peak intensity calculation unit 124C shown in Figure 25, and the process of determining bearing damage according to the peak intensity is performed by the damage detection unit 124D.

[0112] The damage determination procedure described above can be modified as needed. Figure 35 is a flowchart of another damage determination procedure 1, which is a modified version of the flowchart shown in Figure 29. In the other procedure 1 shown in Figure 35, steps S11 to S14 and S18 to S22 described above are common, and the pulse period waveform WF1 generated in S13 is smoothed before being converted to a period ratio to remove errors due to rotational irregularities.

[0113] In other words, the pulse period waveform WF1 generated in S13 is smoothed using the moving average method described above to obtain a pulse period smoothed waveform WF1A (S31). Then, the pulse period smoothed waveform WF1A is divided by the rotation period waveform WF2 generated in S14 to obtain the period ratio waveform WF3A (= WF1A / WF2), which is the period ratio described above, and the period ratio waveform WF3, which is obtained by dividing the pulse period (Tp) waveform WF1 by the rotation period (T) waveform WF2 as described above (S32). Furthermore, the period ratio difference waveform WF5 (= WF3 - WF3A), which is the difference between the period ratio waveform WF3 and the period ratio waveform WF3A, is obtained (S33). Subsequent processing is the same as described above.

[0114] This procedure is the same as the flowchart shown in Figure 29, except that a smoothing process is performed before calculating the period ratio. As in this procedure, the timing of non-dimensionalization as the period ratio is arbitrary, and approximately the same result can be obtained regardless of the timing.

[0115] Figure 36 is a flowchart showing another procedure 2 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. In the other procedure 2 shown in Figure 36, instead of the period ratio smoothing waveform WF4 obtained in S16 as described above, a constant offset value is set as WF4 (S36). This offset value WF4 may be, for example, 1 / number of poles n (n=48) ≈ 0.02083.

[0116] Figure 37 is a flowchart of another procedure 3 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. In the other procedure 3 shown in Figure 37, the period ratio smoothing waveform WF4 obtained in S16 and the period ratio difference waveform WF5 obtained in S17 are omitted, and the period ratio mean distribution WF6, which is the average value of the period ratio for each pole, is obtained from the period ratio waveform WF3 obtained in S15 (S35). Then, the period ratio fluctuation waveform WF7 is obtained by subtracting the value of the period ratio mean distribution WF6 from the period ratio waveform WF3 (S36).

[0117] Figure 38 is a flowchart of another step 4 for damage determination, with some modifications to the steps in the flowchart shown in Figure 29. In the other step 4 shown in Figure 38, the period ratio smoothed waveform WF4 obtained in S16 above is obtained by performing a 7-term moving average process, which is a moving average of the data for 7 poles of the period ratio waveform WF3 (S37). By performing a moving average on terms corresponding to a portion of the total number of poles of the magnetic encoder (n=48), an appropriate noise limiting effect can be obtained. The number of terms in the moving average may be set to, for example, 1 / 5 to 10 of the total number of poles of the magnetic encoder.

[0118] (Second Configuration Example) Figure 39 is a schematic diagram of a second configuration example of the hub unit bearing bearing damage detection system 200. In Figure 39, the hub 115 and sensors 121 and 122 of the hub unit bearing 111A are shown in a horizontal cross-section. The first configuration example shown in Figure 24 is configured to detect the rotation of the hub shaft 131 with one sensor 121, but the second configuration example has the same hub unit bearing configuration as the first configuration example, except that the rotation is detected with two sensors 121 and 122.

[0119] When there is only one sensor, the output signal from the sensor is superimposed with errors due to the displacement of the raceway wheels caused by changes in the position of the rolling elements. Therefore, the fluctuations in each of the waveforms mentioned above include the displacement due to the change in the position of the rolling elements, and peaks caused by this effect appear in the frequency characteristics.

[0120] On the other hand, the bearing damage detection system 400 shown in Figure 39 includes a sensor (first rotation sensor) 121 that detects rotation at one end in the horizontal direction and a sensor 122 (second rotation sensor) that detects rotation at the other end in the horizontal direction, both located at the vertical midpoint of the hub unit bearing 111A. Data processing is performed using the phase difference of the output signals from each of the sensors 121 and 122. Sensor 122 is positioned with its detection surface 122a facing the magnetic encoder 155. The sensor output signals from the two sensors 121 and 122, positioned opposite each other in the horizontal direction, show phase lags or leads in opposite directions due to the displacement of the rolling wheels accompanying changes in the rolling element position. Therefore, by using the difference signal of the sensor output signals from the two sensors 121 and 122, the above-mentioned phase lags or leads are canceled out, and the vertical displacement is calculated. Based on this difference signal (pulse calculation value), data processing is performed to obtain the rotation fluctuation waveform. When the frequency characteristics of the rotational fluctuation waveform are determined, the noise is reduced, and in the case of bearing damage, peaks occurring at a period calculated from the rolling element pitch and the rolling element's orbital speed become clearly visible, and in the case of inner ring damage, peaks occurring at a period calculated from the rotational speed become clear, improving the accuracy of damage detection.

[0121] Figure 40 is a schematic diagram illustrating the arrangement of the inner ring 133, outer ring 113, and magnetic encoder 155 of the hub unit bearing 111A, and the detection area DRa of sensor 121 and the detection area DRb of sensor 122. In Figure 40, the direction of action of the radial load applied from the inner ring 133 to the outer ring 113 (generally the vertical direction Z, which is the direction of gravity) is shown as the up and down direction.

[0122] The positions of the detection region DRa of sensor 121 and the detection region DRb of sensor 122 are not particularly limited, but it is preferable to set the center position Pc of both to a position shifted by a distance δ downward from the horizontal line Lh passing through the central axis O of the hub unit bearing 111A. In this case, as will be described in detail later, it becomes easier to detect the shift using the phase difference signal. Alternatively, the center positions Pc of the detection regions DRa and DRb may be placed on the aforementioned horizontal line Lh.

[0123] Figure 41 is a schematic circuit diagram of the pulse signal generation unit 123A. In this case, the pulse signal generation unit 123A takes the output signal from sensor 121 (sensor A) as pulse signal PL_A and the output signal from sensor 122 (sensor B) as pulse signal PL_B. It also outputs the difference signal between pulse signal A and pulse signal B as a phase difference signal PD (= PL_A - PL_B).

[0124] Figure 42 is an explanatory diagram showing example waveforms of pulse signals PL_A, PL_B, and phase difference signal PD. The pulse period of pulse signal PL_A is TA, and the pulse period of pulse signal PL_B is T Ｂ Therefore, the phase difference between pulse signals PL_A and PL_B is T ＰＤ Therefore, the phase difference signal PD, for example, when the voltages of pulse signals PL_A and PL_B are +V and +2V, will have a waveform that changes in the range of +V to -V, and the period during which it is -V is the phase difference T. ＰＤ It corresponds to this.

[0125] The pulse signals PL_A, PL_B, and phase difference signal PD described above change due to damage to the hub unit bearing 111A. Figure 43 is an explanatory diagram showing the effect of the relative displacement of the inner and outer rings due to the separation of the outer ring raceway on the sensor output. Here, as in the case shown in Figure 27, it shows how the hub shaft 131 moves up and down due to the separation or indentation of the outer ring raceway 127 of the outer row bearing section 149A.

[0126] Assume that a radial load Pr (preload) is applied to the bearing section 149A from the hub shaft 131 to the outer ring 113 in an upward direction in the vertical direction. When a rolling element 117 that is floating in a defective region where separation has occurred in the outer ring raceway 127 rides up from the edge of the defective region to a region without separation, the hub shaft 131 is subjected to the aforementioned impact load, which is greater than the radial load Pr, causing the hub shaft 131 to be displaced downward. Subsequently, if the hub shaft 131 rotates further and an impact load is generated in the opposite direction, the hub shaft 131 is displaced upward. In this case, the magnetic encoder 155 fixed to the hub shaft 131 is displaced in the direction of descending and then rising in accordance with the displacement of the hub shaft 131. Even if the above-described preload is not applied to the hub shaft 131, the behavior of the magnetic encoder 155 is the same as above, and it will descend and rise in accordance with the displacement of the hub shaft 131.

[0127] In other words, in the floating state, the hub shaft 131 is displaced upward, and the center of the outer ring 113 O ｏｕｔ The center of the hub axis 131 relative to the center O ｉｎ The hub shaft shifts upward. At that time, the pulse signals PL_A and PL_B detected from the detection regions DRa and DRb experience a phase lead and lag from the neutral position. Specifically, in detection region DRa, the hub shaft 131 is displaced upward in the same direction as the rotational direction Ro, so the pulse signal PL_A has a phase lead. In detection region DRb, the hub shaft 131 is displaced upward in the opposite direction to the rotational direction Ro, so the pulse signal PL_B has a phase lag. As a result, the phase difference signal PD, which is the difference signal between pulse signal PL_B and pulse signal PL_A, has a phase difference T ＰＤ A period of time will occur.

[0128] On the other hand, when the hub shaft 131 is displaced downward, in detection region DRa the hub shaft 131 is displaced downward in the opposite direction to the rotational direction Ro, causing a phase delay in the pulse signal PL_A. In detection region DRb the hub shaft 131 is displaced downward in the same direction as the rotational direction Ro, causing a phase lead in the pulse signal PL_B. As a result, the phase difference signal PD, which is the difference signal between pulse signal PL_B and pulse signal PL_A, has a different phase difference T than when the hub shaft 131 is displaced upward. ＰＤ A period of time will occur.

[0129] In other words, the phase difference T of the phase difference signal PD ＰＤ Since the value of changes according to the vertical movement of the hub shaft 131, the relative displacement between the inner ring 133 and the outer ring 113 can be calculated from the phase difference of the pulse signals from the two sensors 121 and 122. Phase difference T ＰＤ This changes depending on the position of the detection regions DRa and DRB, and the phase difference T appears in the phase difference signal PD. ＰＤ This can be easily detected by determining the pulse length of a specific pulse. As mentioned above, by setting the center position Pc of the detection regions DRa and DRB to a position shifted below the horizontal line Lh passing through the central axis O of the hub unit bearing 111A, the pulse length of the specific pulse described above can be kept from becoming 0 and thus undetectable.

[0130] The phase difference and pulse period described above are not limited to being calculated from the waveform obtained by the pulse signal generation unit 123A subtracting pulse signals PL_A and PL_B in real time. Alternatively, the waveforms of pulse signals PL_A and PL_B may be monitored individually, and then the phase difference and pulse period may be calculated from each pulse signal.

[0131] The above explanation provides a schematic overview of the signal change behavior; a more accurate description of the signal change behavior is provided below. Figure 44 is an explanatory diagram showing the relationship between the displacement of the magnetic encoder and the phase difference of the detected pulse signal. Figure 45 is an explanatory diagram showing examples of changes in the pulse signals PL_A, PL_B and the phase difference signal PD. Figures 44 and 45 show the displacement when separation occurs in the outer ring raceway as shown in Figure 27, in chronological order in the sections SC1, SC2, SC3, and SC4. In the initial state SC1, the velocity (VR) detected in the detection region DRa and detection region DRb shown in Figure 44 is ０ ,VL ０ The relative velocities (also called the VR) are equal to each other. Then, in SC2, where the hub shaft 131 has rotated, the rolling element reaches the end of the defective region of the outer ring raceway, and the rolling element re-enters the space between the outer ring raceway and the inner ring raceway where there is no separation. At this time, as explained in Figure 27 for the case of a single sensor, the hub shaft 131 is displaced downward. Therefore, the speed (VR) detected in the detection region DRa is equal to each other. １ ) decreases, and the rate detected in detection region DRb (VL １ ) will increase.

[0132] In SC3, where the hub shaft 131 rotates further and the rolling elements transition back to their original rolling state between the outer and inner raceways, the speed (VR) detected in the detection region DRa is ２ ) increases, and the rate detected in detection region DRb (VL １ ) decreases. Then, in SC4, where the hub shaft 131 has rotated further, it returns to the same state as the initial state of SC1.

[0133] The pulse signals PL_A, PL_B, and phase difference signal PD (= PL_A - PL_B) in each section SC1, SC2, SC3, and SC4 generate the distributions of velocity, pulse width of the phase difference signal, and pulse period shown in Figure 45. For example, the vertical displacement can be determined from the pulse width of the phase difference signal, and the displacement velocity can be determined from the pulse periods of pulse signals PL_A and PL_B. Although the actual displacement of the hub shaft 131 occurs between SC1 and SC4, the displacement is detected between SC2 and SC4.

[0134] Figure 46 is an explanatory diagram schematically showing the waveforms of the specific pulse signals PL_A and PL_B and the phase difference signal PD. Pulse signals PL_A and PL_B have a phase delay and lead depending on the height position of the detection regions DRa and DRb. Furthermore, in the aforementioned section SC2, the period of pulse signal PL_A becomes longer and the period of pulse signal PL_B becomes shorter. In section SC3, the period of pulse signal PL_A becomes shorter and the period of pulse signal PL_B becomes longer. Therefore, the phase difference T appears in the phase difference signal PD between sections SC3 and SC4. ＰＤ This phase difference T is shorter than other sections. ＰＤ The changes in this parameter are quantitatively determined during the frequency analysis of the waveform, as described later, and are used to determine the extent of damage.

[0135] (Procedure for damage determination) Figure 47 is a flowchart showing the procedure for determining damage to the hub unit bearing 111A using the two sensors 121 and 122 described above. Each of the following steps is performed based on commands from the control unit 124 shown in Figure 39.

[0136] First, the rotation of the rotationally driven hub unit bearing 111A is detected by sensors 121 and 122, and sensor output signals, which are rotation signals output from sensors 121 and 122, are acquired (S41). These sensor output signals are converted into pulse signals PL_A and PL_B, respectively, by the pulse signal generation unit 123.

[0137] Figure 48 is an explanatory diagram showing the waveforms of pulse signal PL_A and pulse signal PL_B. The control unit 124 controls the pulse period T of pulse signal PL_B. ｉ The phase difference PD between pulse signal PL_B and pulse signal PL_A ｉ The phase difference PDi may be the phase difference signal PD generated by the pulse signal generation unit 123A shown in Figure 39, or it may be the result of individually A / D converting each pulse signal PL_A and PL_B to obtain a phase difference.

[0138] Figure 49 shows the space on the horizontal axis representing the sequence of pulses in the pulse signal PL_B, and the period T on the vertical axis. ｉ and phase difference PD ｉ This is an explanatory diagram showing the changes in the phase difference PD.ｉ The spatial waveform is determined (S42), and damage is determined based on that waveform. In other words, instead of the sensor output signal from one sensor 121, data processing is performed using the phase difference between the sensor output signals from two sensors 121 and 122. The following steps are the same as the first configuration example in terms of basic processing content, with only the "pulse period" being changed to "phase difference," so a detailed explanation of each step is omitted.

[0139] Figure 50 is an explanatory diagram showing an example of a waveform from the pulse output signal to the generation of a rotational fluctuation signal by extracting the rotational fluctuation of the hub shaft 131. The phase difference waveform WF11 corresponds to the phase difference PD in Figure 49. ｉ This is shown (S43), and the rotation period waveform WF12 is a waveform obtained by calculating the change in the period T of one rotation around each pulse of the pulse signal PL_B (which may also be obtained from PL_A) as shown in Figure 8 above (S44).

[0140] Next, the phase difference ratio waveform WF13 (= WF11 / WF12) is obtained by dividing the phase difference waveform WF11 by the rotation period waveform WF12 (S45). In this phase difference ratio waveform WF13, the effect of rotation speed error is removed.

[0141] Then, the phase difference ratio waveform WF13 is smoothed in the same manner as described above to obtain the phase difference ratio smoothed waveform WF14 (S46), and the phase difference ratio difference waveform WF15 (= WF13 - WF14), which represents the difference between the phase difference ratio waveform WF13 and the phase difference ratio smoothed waveform WF14, is obtained (S47). Note that, depending on the conditions, the phase difference ratio smoothed waveform WF14 may not be subtracted from the phase difference ratio waveform WF13, and the phase difference ratio waveform WF13 may be used as the phase difference ratio difference waveform WF15.

[0142] Next, the average phase difference ratio distribution WF16 (reference waveform), which is the average value for each pole (for each rotation angle) of the magnetic encoder, is obtained from the phase difference ratio difference waveform WF15 (S48). Figure 51 is an explanatory diagram showing an example of the average phase difference ratio distribution WF16. In the average phase difference ratio distribution WF16, the deviation from the 0 level on the vertical axis represents the magnetization error of the magnetic encoder. In Figure 51, the line (not shown) connecting the average values ​​for each pole of the average phase difference ratio distribution represents the average phase difference ratio waveform.

[0143] Next, a phase difference ratio fluctuation waveform WF17 (rotation fluctuation waveform) is calculated (S19), which represents the difference between the phase difference ratio difference waveform WF15 and the phase difference ratio average value waveform WF16 described above. The phase difference ratio fluctuation waveform WF17 is obtained by subtracting the values ​​for each pole shown in the phase difference ratio average value waveform WF16 from the values ​​for each pole in the phase difference ratio difference waveform WF15, across the corresponding poles. The phase difference ratio difference waveform WF15 described above corresponds to the "detection waveform," the phase difference ratio average value waveform WF16 corresponds to the "reference waveform," and the phase difference ratio fluctuation waveform WF17 corresponds to the "rotation fluctuation waveform."

[0144] Figure 52 is an explanatory diagram illustrating the phase difference ratio fluctuation waveform WF17. In this phase difference ratio fluctuation waveform WF17, the magnetization error of the magnetic encoder 155 is eliminated. In other words, since the phase difference ratio fluctuation waveform WF17 eliminates rotational speed changes, rotational irregularities, and magnetization errors, the fluctuations that appear here can be attributed to displacements associated with changes in the rolling element positions and displacements associated with damage to the bearing raceway surface if it is damaged.

[0145] Furthermore, the phase difference ratio fluctuation waveform WF17 may be a waveform obtained by converting the vertical axis from the period ratio to the displacement fluctuation using geometric calculations (S50). In that case, the level of fluctuation becomes easier to grasp intuitively as the magnitude of the displacement.

[0146] Next, the phase difference ratio fluctuation waveform WF17 is subjected to frequency analysis (S51). Figure 53 is an explanatory diagram showing an example of the frequency characteristics WF18 obtained by processing the phase difference ratio fluctuation waveform WF17 with FFT. The phase difference ratio fluctuation waveform WF17 includes displacement fluctuations due to changes in the rolling element position, and, if the bearing is damaged, displacement fluctuations due to bearing damage. Therefore, the frequency characteristics WF18 shows spatial frequency peaks (first to severalth order) calculated from the orbital velocity of the balls, peaks that occur with a period calculated from the ball pitch and orbital velocity of the balls if the bearing is damaged, and peaks that occur with a period calculated from the rotational speed if the outer ring is damaged. If such peaks due to bearing damage are above a threshold, it is determined that the bearing is damaged. Note that the same frequency characteristics WF18 can be obtained even if the vertical axis of the phase difference ratio fluctuation waveform WF17 is converted to displacement fluctuation.

[0147] In the frequency response WF18 shown in Figure 53, the main peaks caused by bearing damage appear, for example, as the first-order defect peak Pk1 and the second-order defect peak Pk2. As described above, bearing damage is determined according to the sum of the peak intensities within the set spatial bands BD1 and BD2.

[0148] In addition to the procedure described above, bearing damage may also be determined by the following other procedure. Figure 54 is a flowchart of another procedure 1 for damage determination, which is a modified version of the flowchart shown in Figure 47. In the other procedure 1 shown in Figure 54, the phase difference ratio waveform WF13 obtained in S45 is subjected to a 48-term moving average process, which is performed on the phase difference ratio waveform WF13 by moving the data for all poles (48 poles) of the magnetic encoder to obtain a phase difference ratio smoothed waveform WF14. This phase difference ratio smoothed waveform WF14 is a generally flat waveform and has an approximately constant offset value overall. In S47, the difference between the phase difference ratio waveform WF13 and the phase difference ratio smoothed waveform WF14 is obtained as the phase difference ratio difference waveform WF15.

[0149] Figure 55 is a flowchart of another procedure 2 for damage determination, with some modifications to the steps in the flowchart shown in Figure 47. In the other procedure 2 shown in Figure 55, the phase difference ratio waveform WF13 obtained in S45 is averaged for each pole of the magnetic encoder to obtain the average phase difference ratio distribution WF16 (S52). Then, the value of the average phase difference ratio distribution WF16 is subtracted from the phase difference ratio waveform WF13 to obtain the phase difference ratio fluctuation waveform WF17 (S53). In S51, this phase difference ratio fluctuation waveform WF17 is subjected to frequency analysis.

[0150] Figure 56 is a flowchart of another procedure 3 for damage determination, with some modifications to the steps in the flowchart shown in Figure 47. In the other procedure 3 shown in Figure 56, the phase difference ratio waveform WF13 obtained in S45 is subjected to a 7-term moving average process, which is a moving average of the data for 7 poles of the phase difference ratio waveform WF13, to obtain a phase difference ratio smoothed waveform WF14 (S54). By performing a moving average on terms corresponding to a portion of the total number of poles of the magnetic encoder (n=48), an appropriate noise limiting effect can be obtained. The number of terms in the moving average may be set to, for example, 1 / 5 to 10 of the total number of poles of the magnetic encoder.

[0151] The bearing flaw detection method described above is also applicable to single-row bearings. Furthermore, hub unit bearings used as vehicle support mechanisms are equipped with an encoder, a wheel speed sensor, and a signal processing circuit. In this case, by improving the performance of this hardware, the bearing flaw inspection method described above can be easily implemented without adding a new system, simply by changing the control content, i.e., changing or adding software, and improving the performance of the hardware (improving processing speed, etc.). In addition, in double-row bearings such as hub unit bearings, it is possible to detect damage in either row by providing sensors and encoders in only one row, rather than individually in each row.

[0152] <Alignment Failure Detection System> Next, we will explain the alignment failure detection system for detecting alignment failures. Specific examples of alignment failures include loose fasteners and flange opening. Since these failures exhibit characteristic quantities in the Nth rotation order, these characteristic quantities can be used to detect conditions such as loose fasteners and flange opening on the wheel. Here, the wheel and hub correspond to the flange portions of the rotating bodies mentioned above.

[0153] Figure 57 is an explanatory diagram showing the relative rotation and displacement between the wheel 213 and the hub 225 when the nut 229 loosens during constant-speed rotation, according to the degree of loosening. For example, when changing tires, if the nut 229 is tightened while there is rust on the hub 225, etc., if the nut 229 is tightened to less than the specified tightening torque, or if a different type of nut 229 is used, the nut 229 will loosen. This loosening of the nut 229 can cause phenomena different from the general rotational loosening that occurs when the nut 229 is properly tightened.

[0154] The upper part of Figure 57 is a schematic side view of the wheel 213 and hub 225 as seen from the axial direction, and the lower part is a schematic cross-sectional view of the wheel 213 and hub 225 in the vertical direction. Note that the schematic cross-sectional view in the lower part does not necessarily coincide with the rotational phase of the wheel 213 and hub 225 in the upper part. When the nut 229 is tightened securely to the wheel 213, the wheel 213 and hub 225 are positioned concentrically, and the bolts 227 that attach the wheel 213 to the hub 225 are positioned at the center of each fastening hole 226.

[0155] On the other hand, if all the nuts 229 become slightly loose, the vertical load of the vehicle weight on the hub 225 will cause the spigot portion 228 of the hub 225 to abut against the inner circumferential surface of the central hole 230 of the wheel 213, that is, the gap δ ｈ The hub 225 is displaced downward by 2 units (arrow D1). As the hub 225 is displaced downward, the bolt 227 is pressed downward against the inner circumferential surface of the fastening hole 226 of the wheel 213 at a portion of the fastening hole 226 (position Pc). As a result, the wheel 213 rotates slightly (counterclockwise in Figure 57) while slippage occurs between the inner circumferential surface of the central hole 230 of the wheel 213 and the spigot portion 228, using the pressing force from the bolt 227 at position Pc as a rotational force.

[0156] As the nut 229 loosens further (indicated as "middle" in Figure 57), the rotational force of the hub 225 is transmitted to the wheel 213 through the fastening hole 226 and bolt 227 near region PD of the wheel 213. At the same time, the wheel 213 tilts vertically relative to the hub 225 (opening of the surface indicated by arrow D2). At this point, the entire weight of the wheel is loaded onto the spigot portion 228 of the hub 225.

[0157] Generally, in the case of passenger cars, there is a camber angle θ (the angle of inclination of the tire relative to the vertical direction ND), so the wheel 213 vibrates in the direction of the car's roll (arrow V) while moving in the direction that makes the camber angle θ 0°. Also, when face-opening occurs, the wheel 213 rotates along an elliptical orbit when viewed from the rotation axis direction of the hub 225, and the angular velocity of rotation changes according to the rotation position. In other words, in addition to the wheel 213 rotating along an elliptical orbit, the bolts 227 push the wheel 213 in region PD, so a rotational unevenness equal to the number of bolts occurs for each rotation of the wheel 213.

[0158] If the nut 229 loosens further (indicated as "extra large" in Figure 57), the wheel 213 will detach from the spigot portion 228 of the hub 225 (arrow D3), and the full weight of the wheel will be loaded onto the bolt 227. This will cause even greater vibrations.

[0159] As vibrations occur at each of the above stages, rotational fluctuations may occur in the hub 225 and wheel 213. By detecting these rotational fluctuations, loosening and opening of the nut 229 can be detected. Note that the above-described example of fastening the bolt 227 and nut 229 is just one example; for example, instead of a nut 229, multiple female threads may be formed in the wheel 213, and the male threads of the bolt 227 may be screwed into each female thread.

[0160] Furthermore, the alignment problems described above can take various forms. Figure 58 is an explanatory diagram showing a form of face-opening between rotating bodies. When the wheel 213 side is the second rotating body 19 and the hub 225 side is the first rotating body 17, the face-opening at the joint surface between the wheel 213 and the hub 225 occurs when their respective axes of rotation are tilted relative to each other at an angle φ, and the joint surfaces are fixed in a face-open state. In this form, the axis of rotation of the second rotating body 19 is swung around the axis of rotation of the first rotating body 17.

[0161] Figure 59 is an explanatory diagram showing the form of misalignment between rotating bodies. In this case, the rotation axes of the first rotating body 17 and the second rotating body 19, which are parallel to each other, rotate with a misalignment of distance δ perpendicular to their axes.

[0162] If problems such as face opening or misalignment occur, as illustrated by these examples, the abnormality can be detected quickly by detecting the rotational fluctuations occurring in the second rotating body 19.

[0163] <System for detecting preload loss in rolling bearings> For example, by applying the aforementioned state change detection system to tapered roller bearings used in hub unit bearings and detecting rotational fluctuations, preload loss in the bearings can be detected early. For example, whether it is a tapered roller bearing or a ball bearing used in a hub unit bearing, when preload is applied, the positions of the inner and outer rings are strongly constrained, and the inner and outer rings rotate at almost the same position. However, when preload loss occurs, the constraint between the inner and outer rings weakens, and the relative positions of the inner and outer rings become more prone to fluctuation. Furthermore, when preload loss occurs, the positional relationship between the inner and outer rings changes significantly when the load on the bearing changes, and even if the load on the bearing remains constant, the fluctuation in the positional relationship between the inner and outer rings described above will be greater than when there is no preload loss.

[0164] Figure 60 is a graph showing the distribution of relative changes between the inner and outer rings of a rolling bearing with preload applied. Figure 61 is a graph showing the distribution of relative changes between the inner and outer rings of a rolling bearing with the preload removed. In each graph, the horizontal axis represents displacement in the left-right direction (horizontal direction), and the vertical axis represents displacement in the up-down direction. Displacements in both the left-right and up-down directions can be measured by arranging two sets of sensors, as shown in Figure 39 above: one set of sensors to detect left-right displacement and one set of sensors to detect up-down displacement.

[0165] When preload loss occurs, the variation in relative displacement between the inner and outer rings increases, and consequently, rotational fluctuations also increase. Therefore, by detecting rotational fluctuations using a state change detection system, preload loss in rolling bearings can be detected early.

[0166] <Waveforms of frequency characteristics for each detection target> Next, the state change detection system for the rotating device described above is configured to detect the rotation of the hub unit bearing for the drive wheel of an automobile, and the results of detecting the state change while the automobile is running will be explained. Figure 62 is an explanatory diagram showing the waveforms of frequency characteristics for each detection target of the hub unit bearing. Figure 62 shows the waveforms of frequency characteristics shown in Figures 33 and 53 above, with the case of one sensor as shown in Figure 4 and the case of two sensors as shown in Figure 39.

[0167] In a healthy rotating device, no significant peaks appear in the frequency response waveform, whether with one sensor or two sensors. On the other hand, when indentations occur on the raceway surface of the rolling bearing, or when delamination occurs on the outer ring raceway surface of the rolling bearing, a slight second-order rotational peak appears in the frequency response waveform when there is one sensor, and primary and second-order rotational peaks appear when there are two sensors. By detecting these peaks, changes in condition can be detected.

[0168] In a configuration where the wheel and hub are fastened together with five fasteners (bolts and nuts), if all the nuts loosen by approximately 2 mm in the axial direction, the frequency response waveform shows a fifth-order rotational peak because there are five bolts in total.

[0169] Furthermore, when a spacer was interposed between the wheel and the hub and fastened together with a fastener, causing a gap between the surfaces, a first-order rotational peak appeared in the frequency response waveform.

[0170] When preload loss occurred, the frequency response waveform showed no peaks when using one sensor, but rotational peaks of the 0th to 2nd order appeared when using two sensors. The peak intensity and peak frequency of each of these peaks can be detected using appropriate algorithms. Furthermore, it is possible to detect state changes based on the waveform pattern of the original waveform from which the frequency response was obtained.

[0171] Thus, the present invention is not limited to the embodiments described above. It is also intended that those skilled in the art may modify and apply the embodiments by combining them with each other, based on the description in the specification, and based on well-known technology, and this is within the scope of protection. For example, depending on the type of sensor, the output signal may have waveform distortion, superposition of additional information, etc. Even in such cases, the desired pulse waveform described above can be generated by adding appropriate signal processing, processing circuits, etc. In other words, each of the damage detection steps described above can be performed by appropriately adding the necessary processing according to the waveform of the signal acquired from the sensor.

[0172] As described above, the following matters are disclosed in this specification: (1) A state change detection system for a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: a rotation detection unit that outputs a pulse signal corresponding to the rotation of the rotating body; a pulse calculation unit that obtains a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the pulse signal output from the rotation detection unit; a reference generation unit that generates a reference waveform in which the average pulse calculation waveform obtained by integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body and averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; a rotation fluctuation waveform generation unit that generates a rotation fluctuation waveform by extracting fluctuations other than the steady-state components that occur steadily with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or pulse calculation waveforms continuous with the pulse calculation waveform; and a state change detection unit that detects a change in the rotation state based on the rotation fluctuation waveform. This rotating device state change detection system uses a reference waveform obtained by averaging pulse calculation values ​​for each rotation angle to process the detected pulse signal, thereby obtaining a rotation fluctuation waveform that eliminates the influence of errors and other factors unnecessary for detection. Based on this rotation fluctuation waveform, changes in the rotation state of the rotating body can be reliably detected, contributing to the early detection of state changes.

[0173] (2) The state change detection system for a rotating device according to (1), wherein the pulse calculation waveform is a waveform based on a period ratio waveform that shows the ratio Tp / T of the pulse period Tp of each pulse of the pulse signal to the rotation period T of one rotation of the rotating body centered on the pulse, in the order in which the pulses are generated. According to this state change detection system for a rotating device, by using a period ratio waveform represented by the ratio Tp / T of the pulse period Tp to the rotation period T, the effect of rotation speed error can be removed from the rotation fluctuation waveform.

[0174] (3) The state change detection system for a rotating device according to (2), wherein the pulse calculation waveform is a waveform based on a period ratio difference waveform that represents the difference between the period ratio waveform and the period ratio smoothed waveform obtained by smoothing the period ratio waveform for each rotation angle. According to this state change detection system for a rotating device, by using a period ratio difference waveform that represents the difference between the period ratio waveform and the period ratio smoothed waveform, the effect of rotational fluctuation unevenness error can be removed from the rotational fluctuation waveform.

[0175] (4) The rotational fluctuation waveform generation unit generates the rotational fluctuation waveform by calculating the difference between the values ​​of the same rotation angles of the rotating body in the pulse calculation waveform and the average pulse calculation value of the reference waveform for each rotation angle. This rotational fluctuation waveform generation system generates the rotational fluctuation waveform by calculating the difference between the pulse calculation value and the average pulse calculation value for each rotation angle. This system generates a rotational fluctuation waveform by calculating the difference between the pulse calculation value and the average pulse calculation value for each rotation angle, thereby removing the influence of the rotational position detection error of the rotation sensor from the rotational fluctuation waveform.

[0176] (5) A state change detection system for a rotating device according to any one of (1) to (4), comprising: a frequency analysis unit that determines the frequency intensity distribution of the rotational fluctuation waveform; a peak intensity calculation unit that determines the peak intensity of a specific frequency band from the frequency intensity distribution; and a change detection unit that detects a change in the rotational state from the peak intensity. According to this state change system for a rotating device, a change in the detected rotational state can be appropriately detected by determining the peak intensity of the peaks that appear in the frequency intensity distribution.

[0177] (6) The state change detection system for a rotating device as described in (5), wherein the state change detection unit performs a filtering process to reduce the intensity of a predetermined frequency band from the frequency intensity distribution before determining the peak intensity. With this state change detection system for a rotating device, a waveform can be obtained that allows for easy confirmation of the overall trend of the frequency intensity distribution waveform, confirmation of small fluctuations with waveform undulations removed, and confirmation of fluctuations near the rotation order, etc., through filtering processes such as low-pass filters, high-pass filters, and band-pass filters.

[0178] (7) A state change detection system for a rotating device according to any one of (1) to (6), wherein the rotating body comprises a first rotating body and a second rotating body connected to each other, the first rotating body and the second rotating body each having annular flange portions extending radially opposite to each other, a protruding shaft projecting in the axial direction is provided at the rotation center of one of the pair of flange portions, and a recess is provided on the other of the pair of flange portions for accommodating the protruding shaft, and the pair of flange portions are fastened together by fasteners at multiple locations along their respective circumferential directions with the protruding shaft inserted into the recess.

[0179] (8) The state change detection system for a rotating device as described in (7), wherein the fastener is provided at multiple locations along the circumferential direction of the flange portion on the outer circumference of the pair of flange portions and each fastener has a bolt and a nut that screws onto the bolt or a female threaded portion provided on the flange portion. According to this state change detection system for a rotating device, the pair of flange portions are fastened together by the bolt and nut.

[0180] (9) The rotating device state change detection system according to (8), wherein the first rotating body includes a hub inner ring having an inner ring raceway on its outer circumference and connected to the drive shaft of the vehicle, the second rotating body includes a wheel having the recess and a plurality of fastening holes through which the bolt is inserted, and the rolling bearing is a hub unit bearing comprising the hub inner ring, a hub outer ring having an outer ring raceway on its inner circumference and fixed to the vehicle side, and a plurality of rolling elements provided to be rotatable between the outer ring raceway and the inner ring raceway. According to this rotating device state change detection system, a change in the state of the hub unit bearing of the vehicle can be detected.

[0181] (10) The rotation detection unit comprises a first rotation sensor that detects rotation at one end in the horizontal direction perpendicular to the rotation axis of the rolling bearing, and a second rotation sensor that detects rotation at the other end in the horizontal direction, according to any one of (1) to (9). With this rotation detection system for a rotating device, by using the first rotation sensor and the second rotation sensor in combination, for example, the difference signal of the output signals of each sensor can be obtained to cancel out a phase delay or lead of the signals.

[0182] (11) The state change detection system for a rotating device according to (10), wherein the pulse signal is a signal based on a first pulse signal output from the first rotation sensor and a second pulse signal output from the second rotation sensor, and the pulse calculation unit determines, for each pulse of the first pulse signal or the second pulse signal, a rotation period waveform shown by the progression of the pulse generation order of the pulse signal over the rotation period T of one rotation of the rotating body centered on the pulse, and a phase difference waveform representing the progression of the phase difference PD between the pulses of the first pulse signal and the second pulse signal, and determines a waveform based on a phase difference ratio waveform representing the ratio PD / T of the phase difference PD to the rotation period T for each rotation angle as the pulse calculation waveform.

[0183] (12) The state change detection system for a rotating device as described in (11), wherein the pulse calculation unit obtains a waveform based on a phase difference ratio difference waveform that represents the difference between the phase difference ratio waveform and the phase difference ratio smoothed waveform obtained by smoothing the phase difference ratio waveform, as the pulse calculation waveform. According to this state change system for a rotating device, the effect of rotational unevenness error can be removed from the rotational fluctuation signal by using the phase difference ratio difference waveform.

[0184] (13) A method for detecting a change in the rotation state of a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: outputting a pulse signal corresponding to the rotation of the rotating body; obtaining a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the output pulse signal; integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body; generating a reference waveform in which the average pulse calculation waveform obtained by averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; generating a rotation fluctuation waveform by extracting fluctuations other than the steady component that steadily occurs with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or pulse calculation waveforms continuous with the pulse calculation waveform; and detecting a change in the rotation state based on the rotation fluctuation waveform. According to this method for detecting a change in the rotation device, a rotation fluctuation waveform can be obtained that eliminates the influence of errors and the like that do not need to be detected by processing the detected pulse signal using a profile waveform obtained by averaging the pulse calculation values ​​for each rotation angle. Based on this rotational fluctuation waveform, changes in the rotational state of a rotating body can be reliably detected, contributing to the early detection of state changes.

[0185] (14) A method for detecting a change in the state of a rotating device according to (13), which detects at least one of the following based on the rotational fluctuation waveform: damage to the raceway surface or rolling surface of the rolling bearing, loosening of fasteners connecting a plurality of rotating bodies, inclination of the rotation axes of the rotating bodies, or loss of preload in the rolling bearing. This method for detecting a change in the state of a rotating device makes it possible to reliably and early detect at least one of the following: damage to the raceway surface or rolling surface of the rolling bearing, loosening of fasteners, inclination of the rotation axes of the rotating bodies, or loss of preload in the rolling bearing.

[0186] This application is based on Japanese Patent Application No. 2024-176653 filed on October 8, 2024, the contents of which are incorporated herein by reference.

[0187] 11 Rotating device 12 Rotating body 13, 119 Rotation detection unit 15, 24, 124 Control unit 17 First rotating body 19 Second rotating body 21 Fastener 21A Bolt 21B, 229 Nut 23, 25 Rolling bearing 23A, 25A, 133 Inner ring 23B, 25B, 113 Outer ring 27, 31 Shaft body 29, 33 Flange portion 35 Through hole 37, 38 Protruding shaft 39 Recess 43, 121, 122 Sensor 45, 123, 123A Pulse signal generation unit 47, 124A Rotation fluctuation extraction unit 49 Pulse calculation unit 51 Reference generation unit 53 Rotation fluctuation waveform generation unit 55 State change detection unit 57, 155 Magnetic encoder 57 Magnetic encoder 100 State change detection system 111, 111A Hub unit bearing 115, 225 Hub 117, 117A Rolling elements 121a, 122a Detection surface 124B Frequency analysis unit 124C Peak intensity calculation unit 124D Damage detection unit 125 Stationary flange 127, 129 Outer ring raceway 131 Hub shaft 135 Mounting flange 135a Through hole 137 Hub bolt 139, 145 Inner ring raceway 141 Small diameter step 143 Crimping part 149A, 149B Bearing part 151 Seal ring 153 Internal space 155a Support ring 155b Encoder body 159 Side cover 200, 300, 400 Bearing damage detection system 213 Wheel 226 Fastening hole 227 Bolt 228, spigot portion 230, center hole

Claims

1. A state change detection system for a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: a rotation detection unit that outputs a pulse signal corresponding to the rotation of the rotating body; a pulse calculation unit that obtains a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the pulse signal output from the rotation detection unit; a reference generation unit that generates a reference waveform in which the average pulse calculation waveform obtained by integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body and averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; a rotation fluctuation waveform generation unit that generates a rotation fluctuation waveform by extracting fluctuations other than the steady component that steadily occurs with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or a pulse calculation waveform continuous with the pulse calculation waveform; and a state change detection unit that detects a change in the rotation state based on the rotation fluctuation waveform.

2. The state change detection system for a rotating device according to claim 1, wherein the pulse calculation waveform is a waveform based on a period ratio waveform that shows the ratio Tp / T of the pulse period Tp of each pulse of the pulse signal to the rotation period T of one rotation of the rotating body centered on the pulse, in the order in which the pulses of the pulse signal are generated.

3. The state change detection system for a rotating device according to claim 2, wherein the pulse calculation waveform is a waveform based on a period ratio difference waveform that represents the difference for each rotation angle between the period ratio waveform and the period ratio smoothed waveform obtained by smoothing the period ratio waveform.

4. The rotational fluctuation waveform generation unit generates the rotational fluctuation waveform by calculating the difference between the values ​​of the same rotation angles of the rotating body in the pulse calculation waveform and the average pulse calculation value of the reference waveform for each rotation angle. This is the state change detection system for a rotating device according to claim 1.

5. The state change detection system for a rotating device according to claim 1, comprising: a frequency analysis unit for determining the frequency intensity distribution of the rotational fluctuation waveform; a peak intensity calculation unit for determining the peak intensity of a specific frequency band from the frequency intensity distribution; and a change detection unit for detecting a change in the rotational state from the peak intensity.

6. The state change detection system for a rotating device according to claim 5, wherein the state change detection unit performs a filtering process to reduce the intensity of a predetermined frequency band from the frequency intensity distribution before determining the peak intensity.

7. The rotating body comprises a first rotating body and a second rotating body connected to each other, the first rotating body and the second rotating body each having annular flange portions extending radially opposite to each other, a projection shaft projecting axially is provided at the rotation center of one of the pair of flange portions, and a recess for accommodating the projection shaft is provided at the other of the pair of flange portions, the pair of flange portions are fastened together at multiple locations along their respective circumferential directions by fasteners with the projection shaft inserted into the recess, the state change detection system for a rotating device according to claim 1.

8. The state change detection system for a rotating device according to claim 7, wherein the fastener is provided at multiple locations along the circumferential direction of the flange portion on the outer circumference of a pair of flange portions, and each fastener has a bolt and a nut that screws onto the bolt or a female threaded portion provided on the flange portion.

9. The state change detection system for a rotating device according to claim 8, wherein the first rotating body includes a hub inner ring having an inner ring raceway on its outer circumference and connected to the drive shaft of a vehicle, the second rotating body includes a wheel having the recess and a plurality of fastening holes through which the bolt is inserted, and the rolling bearing is a hub unit bearing comprising the hub inner ring, a hub outer ring having an outer ring raceway on its inner circumference and fixed to the vehicle side, and a plurality of rolling elements provided to be rotatable between the outer ring raceway and the inner ring raceway.

10. The rotation detection unit comprises a first rotation sensor that detects rotation at one end in the horizontal direction perpendicular to the rotation axis of the rolling bearing, and a second rotation sensor that detects rotation at the other end in the horizontal direction, as described in claim 1.

11. The pulse signal is a signal based on a first pulse signal output from the first rotation sensor and a second pulse signal output from the second rotation sensor, and the pulse calculation unit determines, for each pulse of the first pulse signal or the second pulse signal, a rotation period waveform shown by the progression of the pulse generation order of the pulse signal over a rotation period T of the rotating body centered on the pulse, and a phase difference waveform representing the progression of the phase difference PD between the pulses of the first pulse signal and the second pulse signal, and determines a waveform based on a phase difference ratio waveform representing the ratio PD / T of the phase difference PD to the rotation period T for each rotation angle as the pulse calculation waveform, the state change detection system for a rotating device according to claim 10.

12. The state change detection system for a rotating device according to claim 11, wherein the pulse calculation unit determines a waveform based on a phase difference ratio difference waveform that represents the difference between the phase difference ratio waveform and a phase difference ratio smoothed waveform obtained by smoothing the phase difference ratio waveform, and uses this waveform as the pulse calculation waveform.

13. A method for detecting a change in the rotation state of a rotating device having a rotating body and a rolling bearing that rotatably supports the rotating body, comprising: outputting a pulse signal corresponding to the rotation of the rotating body; obtaining a pulse calculation waveform representing the pulse period for each rotation angle of the rotating body from the output pulse signal; integrating the values ​​of the pulse calculation waveform for each rotation angle of the rotating body; generating a reference waveform in which the average pulse calculation waveform obtained by averaging the integrated values ​​obtained for each rotation angle is represented as a distribution for one rotation of the rotating body; generating a rotational fluctuation waveform by extracting fluctuations other than the steady component that steadily occurs with the rotation of the rotating body for each rotation angle using the reference waveform from the pulse calculation waveform used to generate the reference waveform or pulse calculation waveforms continuous with the pulse calculation waveform; and detecting a change in the rotation state based on the rotational fluctuation waveform.

14. A method for detecting a change in the state of a rotating device according to claim 13, which detects, based on the rotational fluctuation waveform, at least one of the following: damage to the raceway surface or rolling surface of the rolling bearing, loosening of fasteners connecting a plurality of rotating bodies, inclination of the rotation axes of the rotating bodies, or loss of preload of the rolling bearing.

Images