JP3824546B2 - Rotational accuracy measuring device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a rotational accuracy measuring device that measures rotational accuracy of a rotating body that rotates about a predetermined axis.
[0002]
[Prior art]
In a rotational accuracy measuring device that measures the rotational accuracy of a rotating body (for example, a spindle that is rotated by a motor) that is supported by a rolling bearing or a fluid bearing and rotates, time series data indicating the displacement of the rotating body is conventionally used to rotate the rotating body. A large amount of displacement data was collected regardless of the amount of data collected, and data processing was collectively performed for the large amount of displacement data. In this data processing, for example, fast Fourier transform (hereinafter referred to as “FFT”) is performed on the displacement data (hereinafter referred to as “collected data”) obtained in time series, whereby data indicating a frequency spectrum (hereinafter referred to as “FFT”). By calculating the spectral data (hereinafter referred to as “spectrum data”) and removing the synchronous shake error (RRO: Repeatable Run Out) in the spectral data, the asynchronous shake error (NRRO: Non) corresponding to the displacement of the rotating body that is not repeated every rotation. Repeatable Run Out).
[0003]
[Problems to be solved by the invention]
However, in the conventional rotational accuracy measuring device, rotation unevenness of a driving source such as a motor that rotates a rotating body has a great influence on the measurement. Further, since the collected data is time-series data collected regardless of the rotation of the rotating body, when the FFT is performed in the data processing, the collected data is smoothly set to 0 outside a predetermined interval. As shown, appropriate weighting is performed on the collected data by a window function in advance. In general, when a window function is used, the spectrum obtained by the FFT is diffused. Therefore, the use of this window function also greatly affects the measurement by the conventional rotational accuracy measuring device. As described above, in the conventional measuring apparatus, since the influence of the rotation irregularity of the motor and the window function is large, it is difficult to accurately measure the rotation accuracy, that is, to accurately calculate RRO, NRRO, roundness, and the like. .
[0004]
On the other hand, the applicant of the present application generates block data by separating the displacement data as time-series data consisting of sampled values indicating the displacement of the rotating body for each rotation period, and blocking the data. Development of a rotational accuracy measurement device based on periodic separation type signal processing, in which the number of sampling values constituting each data block in the same is made the same by rate conversion, and NRRO and the like are calculated using the blocked data after rate conversion ing. According to this period-separated signal processing, the influence on measurement due to rotation unevenness, use of a window function, etc. is reduced, and accurate calculation of NRRO and the like becomes possible.
[0005]
In the periodic separation type signal processing described above, when a signal synchronized with the rotation period (for example, a signal in which one pulse appears for each rotation) is output as an index pulse from a drive unit including a motor that rotates a rotating body. The index pulse can be used to separate the displacement data for each rotation period. On the other hand, when a signal synchronized with the rotation cycle such as an index pulse cannot be obtained, it is necessary to separate the displacement data for each rotation cycle, for example, by detecting the zero point after removing the DC component from the displacement data. There is. However, it is difficult to accurately determine the zero point due to disturbances such as noise and errors of the A / D converter, and as a result, there is a problem that the displacement data cannot be separated accurately for each rotation period. It was.
[0006]
Therefore, an object of the present invention is to provide a rotation accuracy measuring device capable of accurately separating displacement data for each rotation cycle for accurate rotation accuracy measurement.
[0007]
[Means for Solving the Problems and Effects of the Invention]
A first invention is a rotational accuracy measuring device for measuring rotational accuracy of a rotating body that rotates about a predetermined axis,
Data acquisition means for acquiring, as displacement data, time-series data consisting of sampling values of signals indicating displacement in the radial direction of the rotating body or displacement in the predetermined axis direction;
A moving average means for calculating a moving average for each sampling value constituting the displacement data, and generating smoothed displacement data that is time-series data composed of the moving average;
Blocking means for separating and blocking the displacement data acquired by the data acquisition means for each rotation period of the rotating body based on the zero point in the signal waveform represented by the smoothed displacement data;
Conversion means for performing rate conversion on each data block so that the number of sampling values constituting each data block in the displacement data blocked by the blocking means is the same;
Calculation means for calculating an index indicating the rotation accuracy based on the blocked data composed of the data blocks after the rate conversion ;
The blocking means includes
Based on the smoothed displacement data, the moving average at one sampling point is a positive value and the moving average at the other sampling point is a negative value among the adjacent sampling point pairs at the sampling point of the sampling value indicating the displacement of the rotating body. Sampling point pair detection means for detecting adjacent sampling point pairs,
Separating means for separating the displacement data acquired by the data acquiring means for each rotation period on the assumption that the zero point exists between the detected pair of adjacent sampling points .
[0008]
According to the first aspect of the invention, the displacement data representing the displacement of the rotating body is separated and blocked for each rotation period, and an index indicating the rotation accuracy is calculated based on the blocked displacement data. In such processing, since the separation of the displacement data for each rotation period is performed based on the smoothed displacement data consisting of the moving average for each sampled value in the displacement data, the displacement data is accurately separated for each rotation period. be able to. At this time, among the adjacent sampling point pairs at the sampling point of the sampling value indicating the displacement of the rotating body, an adjacent sampling point pair that can be regarded as having a zero point between them is detected, and the displacement data is based on the adjacent sampling point pair. Separated every rotation cycle. Therefore, it is possible to efficiently and accurately separate the displacement data for each rotation period without obtaining an accurate zero point position. In this way, the displacement data is accurately separated for each rotation period and is divided into blocks, and further rate conversion is performed so that the number of sampling values constituting each data block is the same. Then, an index indicating the rotation accuracy is calculated based on the blocked data after the rate conversion. Therefore, in the rotation accuracy measurement, it is possible to reduce the influence of rotation unevenness such as a motor that rotates the rotating body. Further, when the displacement data is blocked for each rotation period, the synchronous component and its harmonic component in the displacement of the rotating body are obtained by performing FFT on the blocked displacement data without using a window function. Can be obtained as a line spectrum without spread. For this reason, when obtaining the NRRO from the spectrum data obtained by the FFT, the RRO can be accurately removed without affecting the other.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.
<Configuration and operation>
FIG. 1 is a configuration diagram schematically showing a configuration of a rotational accuracy measuring device according to an embodiment of the present invention, and FIG. 2 is a plan view showing measurement of a rotating body by the rotational accuracy measuring device.
[0014]
This rotational accuracy measuring device is a device that measures the rotational accuracy of a rotating body 10 supported by a bearing and rotated by a motor (not shown) around a predetermined rotating shaft 11. Based on the non-contact type displacement sensor 20 to be detected and the sampled value indicating the displacement in the radial direction of the rotating body 10 based on the detection signal Sd of the displacement sensor 20, time series data is collected, and the collected data which is the time series data is collected. And a data collection processing device 30 for processing.
[0015]
The displacement sensor 20 is disposed in the vicinity of the outer peripheral surface of the rotator 10, detects the distance between the outer peripheral surface and the displacement sensor 20, and outputs the detection result as a detection signal Sd indicating the radial displacement of the rotator 10. . When detecting the axial displacement of the rotating body 10 and measuring the rotational accuracy of the axial displacement, the axis of the rotating body 10 is located near the upper surface of the rotating body 10 as shown by the dotted line in FIG. If a non-contact type displacement sensor 21 for detecting the displacement in the direction is arranged and a signal output from the displacement sensor 21, that is, a signal indicating the distance between the upper surface of the rotating body 10 and the displacement sensor 21 is used as the detection signal Sd. Good. However, in the following, for convenience of explanation, it is assumed that only the displacement sensor 20 for detecting the displacement in the radial direction is arranged.
[0016]
The data collection processing device 30 has a configuration in which a CPU 31 as a central processing unit, an input interface unit 32, a memory 33, and a display control unit 34 are connected by a bus, and a display unit 36 is connected to the display control unit 34. ing. The detection signal Sd from the displacement sensor 20 is input to the input interface unit 32 in the data collection processing device 30. The input interface unit 32 includes an A / D converter, and the detection signal Sd is sampled by the A / D converter and temporarily stored as digital data in the memory 33 (hereinafter, this digital data is referred to as digital data). "Original data"). The input interface unit 32, together with the displacement sensor 20, constitutes time series data acquisition means representing the displacement of the rotating body 10.
[0017]
The CPU 31 sequentially executes data processing such as DC cut processing, period separation processing, rate conversion, and FFT, which will be described later, on the original data by executing a predetermined program stored in the memory 33 in advance. Accordingly, the data collection processing device 30 has a functional configuration as shown in FIG. 3, that is, a device including a DC cut unit 111, a period separation unit 112, a rate conversion unit 113, and a signal processing unit 114. Works as.
[0018]
The DC cut unit 111 removes a direct current component from the detection signal Sd output from the displacement sensor 20, and specifically, the detection signal Sd is performed by signal processing on original data that is digital data representing the detection signal Sd. The digital data representing the signal from which the DC component is removed from is generated as the collected data Da (this signal processing is referred to as “DC cut processing”). Although the DC cut unit 111 is realized by software, a DC component cutoff circuit or an offset adjustment circuit is provided in the input interface unit 32, thereby removing the DC component from the detection signal Sd, and then A / D conversion. The sampling data Da may be created by a container. In this case, the DC cut unit 111 is realized as hardware, and constitutes a part of the input interface unit 32.
[0019]
The period separation unit 112 generates block data Db composed of a plurality of data blocks by separating the collection data Da generated by the DC cut unit 111 into blocks for each rotation period of the rotating body 10 (this is the block data Db). Processing is called “periodic separation processing”). Specifically, from the collection data Da that is digital data representing the detection signal Sd, the collection data Da is separated into blocks based on the zero point, which is the time when the displacement of the rotating body 10 becomes zero, and is blocked. Data Db is obtained.
[0020]
FIG. 4 is a flowchart showing the procedure of the period separation process. In this embodiment, when the CPU 31 executes these procedures, the period separation unit 112 is realized by software.
[0021]
In the periodic separation process, the CPU 31 first calculates a moving average for each sampled value constituting the collection data Da as displacement data output from the DC cut unit 111 (step S10). That is, paying attention to each sampled value in the sampling data Da in sequence, an average value of sampled values of a plurality of sampling points in a predetermined period centering on a sampling point of the sampled value of interest (hereinafter referred to as “targeted sampling point”) is obtained. The average value is a moving average for the sampled value of interest. In this way, time-series data consisting of moving averages for each sampled value in the collected data Da is obtained (hereinafter, this time-series data is referred to as “smoothed displacement data”).
[0022]
When the sampling data Da representing the signal waveform as shown in FIG. 5 is obtained from the detection signal Sd indicating the displacement of the rotating body 10, when a part of the signal waveform is enlarged, the signal waveform is, for example, FIG. As shown in (), a smooth waveform including a noise component that fluctuates irregularly is not obtained. Further, the collected data Da, which is digital data obtained from a signal having such a waveform, includes an error caused by an A / D converter (hereinafter referred to as “A / D conversion error”). However, if the above-described moving average processing is performed, noise and A / D conversion errors are canceled out and removed, and digital data representing a smoothed waveform signal as shown in FIG. 6B is smoothed. Obtained as converted displacement data. Note that vertical lines drawn at regular intervals in FIGS. 5 and 6 correspond to sampling points.
[0023]
When such smoothed displacement data is obtained, the CPU 31 detects a zero point in the signal waveform shown in FIG. 6B represented by the smoothed displacement data (step S12). Here, the zero point includes a zero point when the signal changes from a negative value to a positive value (hereinafter referred to as “rising zero point”) and a zero point when the signal changes from a positive value to a negative value ( Hereinafter referred to as “falling zero point”). In order to separate the collection data Da for each rotation period, only one of these two types of zero points may be detected. However, in the following, description will be made assuming that only the rising zero point is detected. Further, in order to separate the collection data Da for each rotation cycle, it is not necessary to accurately obtain the zero point, and two adjacent sampling points may be obtained with the zero point interposed therebetween. Therefore, in the present embodiment, the second sampling point pair is a sampling point pair including a first sampling point and a second sampling point that follows the first sampling point, and the moving average with respect to the sampling value of the first sampling point is a negative value. A sampling point pair (hereinafter referred to as “rising zero-crossing sampling point pair”) whose moving average with respect to the sampled value of the point is a positive value is obtained. The rising zero-crossing sampling point pair can be easily obtained by sequentially examining the positive and negative of the moving average at two adjacent sampling points in the smoothed displacement data.
[0024]
In the example shown in FIG. 6B, the sampling point pair (ta1, ta2) and the sampling point pair (tc1, tc2) are detected as rising zero-crossing sampling point pairs. In this case, the rising zero point Zpa exists between the rising zero crossing sampling point pair (ta1, ta2), and the next rising zero point Zpc exists between the rising zero crossing sampling point pair (tc1, tc2). . Therefore, the sampling points (including ta2 and tc1) from the sampling points ta2 to tc1 correspond to one rotation period of the rotating body 10. The sampling point pair (tb1, tb2) is a falling zero crossing sampling point pair, and a falling zero point Zpb exists between them. Although the sampling data Da can be separated for each rotation cycle by detecting such a falling zero point, as described above, only the rising zero point pair is detected in the present embodiment.
[0025]
Next, the CPU 31 separates the collected data Da by the detected rising zero point and blocks it for each rotation cycle, thereby obtaining the blocked data Db composed of a plurality of data blocks (step S14). In the example shown in FIG. 6B, the rising zero point Zpa is between the rising zero-crossing sampling point pair (ta1, ta2), and the next rising edge is between the rising zero-crossing sampling point pair (tc1, tc2). Since the zero point Zpc exists, the sampling value before the sampling point ta1 and the sampling value after the sampling point ta2 are separated, and the sampling value before the sampling point tc1 and the sampling value after the sampling point tc2 are separated. And sampling values of sampling points (including ta2 and tc1) from sampling points ta2 to tc1 are blocked as one data block.
[0026]
When the period separation process as described above is applied to the sampling data Da shown in FIG. 5, as shown in FIG. 7, each block is made up of four block data Db1 to Db4 corresponding to displacement data for one rotation period. Data Db is obtained. The number of sampled values constituting each data block in the block data Db thus obtained (this is the number of sampling points in one rotation cycle, hereinafter referred to as “data number”) is due to uneven rotation of the motor. , Usually not all equal. For example, as shown in FIG. 8A, the data numbers of the data blocks Db1, Db2, Db3, and Db4 are n1, n2, n3, and n4, respectively.
[0027]
The rate conversion unit 113 makes the number of data of each data block the same by performing interpolation processing on the blocked data Db composed of a plurality of data blocks in which the number of data varies as described above. That is, the number of sampling points in each data block is made the same by rate conversion. At this time, considering the FFT (Fast Fourier Transform) executed by the signal processing unit 114, the number of data in each data block is set to a power of two. For example, by performing rate conversion on the blocked data Db as shown in FIG. 8 (a), data blocks Dc1, Dc2, Dc3, all of which are 2 ^m in number as shown in FIG. 8 (b). Blocked data Dc consisting of Dc4 is obtained. Here, the four data blocks Db1, Db2, Db3, Db4 constituting the blocked data Db are converted into the four data blocks Dc1, Dc2, Dc3, Dc4 constituting the blocked data Dc, respectively. Shall.
[0028]
The signal processing unit 114 is a means for calculating an index indicating rotation accuracy, and calculates spectrum data by performing FFT on the blocked data Dc after rate conversion without using a window function. And based on this spectrum data, RRO, NRRO, roundness, etc. are calculated | required by the method similar to the past.
[0029]
Indices indicating the rotational accuracy of the rotating body 10 such as RRO, NRRO, and roundness obtained in this manner are stored in the memory 33 as measurement results, and based on other predetermined programs, the display control unit 34. And is displayed on the display unit 36 by the display control unit 34.
[0030]
<Effect>
According to the present embodiment as described above, the sampling data Da is separated for each rotation period, whereby the blocked data Db that is blocked for each rotation period is obtained, and the rate conversion for the blocked data Db is further performed. Thus, the number of data in each data block becomes the same. Thereby, in rotation accuracy measurement, the influence of the rotation nonuniformity of the motor which rotates the rotary body 10 can be reduced. For this reason, it is possible to accurately obtain an index indicating rotational accuracy, such as RRO, NRRO, and roundness. Further, by applying FFT to the data separated for each rotation period and made into a block without using a window function, the harmonics of the detection signal Sd having a fundamental frequency that is the reciprocal of one rotation period of the rotating body 10. The component can be obtained as a line spectrum having no spread. For this reason, RRO can be accurately obtained from the spectrum data obtained by performing the FFT, and when obtaining NRRO from the spectrum data, RRO can be accurately removed without affecting the other. it can.
[0031]
By the way, in the present embodiment, in the period separation process, separation of the collection data Da for each rotation period is performed based on zero point detection. At that time, the zero point detection includes moving average for each sampling value of the collection data Da. Smoothed displacement data consisting of For this reason, the collection data Da can be accurately separated for each rotation cycle. That is, if the collected data Da is used as it is for zero point detection, a plurality of zero points may apparently exist in the vicinity of one zero point (true zero point) due to noise, A / D conversion error, or the like. However, by using the smoothed displacement data, the zero point can be accurately detected in a one-to-one correspondence with the original zero point. For example, when a zero point is detected from the collected data Da shown in FIG. 6A, a rising zero point is detected in two sections between the sampling point pair (ta1, ta2) and the sampling point pair (ta2, ta3). There may be instances where a rising zero crossing sampling point pair that exists in appearance and includes a true rising zero point cannot be detected. For this reason, the collection data Da cannot be accurately separated for each rotation cycle. On the other hand, when the zero point is detected from the smoothed displacement data shown in FIG. 6B, it is detected that the rising zero point Zpa exists between the sampling point pair (ta1, ta2). No zero point is detected. For this reason, it becomes possible to isolate | separate the collection data Da correctly for every rotation period.
[0032]
As can be seen from the above-described procedure of the periodic separation process, it is not necessary to obtain the exact position of the zero point in order to separate the collection data Da, and it is sufficient to obtain the sampling point pair in which the zero point exists. . That is, if either one of the rising zero crossing sampling point pair or the falling zero crossing sampling point pair is detected, it is assumed that there is a zero point between the detected sampling point pairs, and the collected data Da is separated for each rotation period. can do. In the periodic separation process, for each sampled value of the sampled data Da, an average of the sampled values of the sampling points for a predetermined period centered on the sampling points of the sampled values is calculated. Must be set appropriately. In the present embodiment, the zero point corresponding to the original zero point is smoothed displacement data in consideration of the rotation period of the rotating body 10, the sampling frequency, the noise level, the frequency, and the like in this predetermined period. An appropriate predetermined period shall be selected so as to be detected from
[0033]
<Modification>
In the above embodiment, rate conversion is performed in order to make the number of data of each data block obtained by the period separation process the same, but instead, the following process may be performed. That is, the number of data with the highest appearance frequency is obtained from the number of data of each data block obtained by separating the collected data Da for each rotation period by the period separation process, and only the data block having the number of data with the highest appearance frequency is obtained. Extracted from the collected data Da. Then, an index indicating the rotation accuracy, such as NRRO, is calculated using the block data composed of the data blocks extracted in this way. Such a modification can also reduce the influence of rotation unevenness of the motor that rotates the rotating body 10 in the rotation accuracy measurement.
[Brief description of the drawings]
FIG. 1 is a configuration diagram schematically showing a configuration of a rotational accuracy measuring apparatus according to an embodiment of the present invention.
FIG. 2 is a plan view showing measurement of a rotating body by the rotation accuracy measuring device according to the embodiment.
FIG. 3 is a block diagram showing a functional configuration of a data processing device in the rotational accuracy measuring device according to the embodiment.
FIG. 4 is a flowchart showing a procedure of period separation processing in the embodiment.
FIG. 5 is a signal waveform diagram for explaining collected data in the embodiment.
FIG. 6 is a signal waveform diagram for explaining period separation processing in the embodiment.
FIG. 7 is a signal waveform diagram for explaining blocked data obtained by a periodic separation process on collected data in the embodiment.
FIG. 8 is a data configuration diagram for explaining rate conversion in the embodiment.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... Rotating body 11 ... Rotating shaft 20, 21 ... Displacement sensor 30 ... Data collection processing apparatus 31 ... CPU
32 ... Input interface unit 33 ... Memory 111 ... DC cut unit 112 ... Period separation unit 113 ... Rate conversion unit 114 ... Signal processing unit Sd ... Detection signal Da ... Collection data (displacement data)
Db ... Blocked data Dc ... Blocked data after rate conversion

Claims (1)

A rotational accuracy measuring device that measures rotational accuracy of a rotating body that rotates about a predetermined axis,
Data acquisition means for acquiring, as displacement data, time-series data consisting of sampling values of signals indicating the radial displacement of the rotating body or the displacement in the predetermined axial direction;
A moving average means for calculating a moving average for each sampling value constituting the displacement data, and generating smoothed displacement data that is time-series data composed of the moving average;
Blocking means for separating and blocking the displacement data acquired by the data acquisition means for each rotation period of the rotating body based on the zero point in the signal waveform represented by the smoothed displacement data;
Conversion means for performing rate conversion on each data block so that the number of sampling values constituting each data block in the displacement data blocked by the blocking means is the same;
Calculation means for calculating an index indicating the rotation accuracy based on the blocked data composed of the data blocks after the rate conversion ;
The blocking means includes
Based on the smoothed displacement data, the moving average at one sampling point is a positive value and the moving average at the other sampling point is a negative value among the adjacent sampling point pairs at the sampling point of the sampling value indicating the displacement of the rotating body. Sampling point pair detection means for detecting adjacent sampling point pairs,
Rotational accuracy , comprising separation means for separating the displacement data acquired by the data acquisition means for each rotation period on the assumption that the zero point exists between the detected pair of adjacent sampling points measuring device.

Images