Rotational-device method and system

• US 5,734,579 A
• Filed: 08/18/1995
• Issued: 03/31/1998
• Est. Priority Date: 09/07/1994
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Sakaguchi, Yoshitami
• Primary Examiner(s): Voeltz, Emanuel T.
• Assistant Examiner(s): VO, HIEN XUAN
• Application Number:
  US08/516,839
• Time in Patent Office: 956 Days
• Field of Search: 364/487, 364/508, 395/183.18, 73/659, 73/660, 73/602, 360/97.01, 360/78.06, 324/76.12, 324/76.13, 324/76.19
• US Class Current: 702/78
• CPC Class Codes: G01H 3/00 Measuring characteristics of vibrations by using a detector in a fluid G01H7/00, G01H9/00, G01H11/00 take precedence

Abstract

A rotational-device test system and method is provided to classify a rotational device according to the type of drive sound thereof. A divider divides the time-series data of an amplitude of the drive sound of an HDD digitized in a converter at intervals of a predetermined time and generates a time-series data group for each frame. An analyzer converts each data item of the time-series data group to a power spectrum. A cycle arithmetic unit selects a maximum point whose self-correlation between a leading frame and each of frames other than the leading frame is maximum, as a candidate point for one cycle of the drive sound, and detects a point whose self-correlation is maximum and which has as a starting point the leading time-series data in the leading frame and the candidate point, and, based on the frame of the candidate point in the detected point whose self-correlation is maximum and on the leading frame, specifies the cycle of the drive sound. A comparator compares a sample having a cycle substantially identical to the cycle of the drive sound of the HDD stored in a storage device and the power spectrum of one cycle by a mutual correlation, and a decision device classifies the HDD based on the result of the comparison.

Claims

1. A rotational-device test system comprising:

a detection means for detecting time-series data on the amplitude of a drive sound of a rotational device;

a digital conversion means for converting the detected time-series data to digital data;

a power spectrum conversion means for converting a plurality of frame time-series data, obtained by dividing the converted digital time-series data into a plurality of frames at intervals of a predetermined time, to a plurality of power spectra through frequency conversion;

a cycle specification means for determining a candidate point, which becomes the candidate for the starting point of a cycle of the drive sound, for each of said frames, and for specifying said cycle of the drive sound by specifying said starting point of the cycle from the candidate points of said plurality of frames; and

a rotational-device classification means for classifying said rotational device based on the power spectrum of one cycle of the drive sound of said rotational device which is based on the specified cycle and the converted power spectrum.

2. The rotational-device test system as set forth in claim 1, wherein said candidate point is specified by a self-correlation.

3. The rotational-device test system as set forth in claim 1, wherein said starting point of a cycle is specified by a self-correlation.

4. The rotational-device test system as set forth in claim 1, wherein said cycle specification means selects a maximum point whose self-correlation between a leading frame of said plurality of frames and each of the other frames other than said leading frame is maximum, as a candidate point for determining said one cycle of the drive sound, and detects a point whose self-correlation is maximum, the self-correlation having as a starting point the leading time-series data in said leading frame and said candidate point, and based on the frame of said candidate point in the detected point whose self-correlation is maximum and on said leading frame, specifies said cycle of said drive sound.

5. The rotational-device test system as set forth in claim 4, wherein said maximum point is obtained by detecting a point whose self-correlation point is maximum, the self-correlation point having said leading time-series data in said leading frame and leading time-series data in frames other than said leading frame, as a starting point, for every combination of said leading frame and each of the frames other than said leading frame.

6. The rotational-device test system as set forth in claim 1, further comprising a sample storage means for storing, in advance, a sample of the power spectrum of one cycle of the drive sound of each of the rotational devices according to the types of rotational devices, and wherein said rotational-device classification means classifies said rotational device by selecting samples having a cycle substantially equal to said cycle of the drive sound of the rotational device from the samples stored in said sample storage means and by detecting a sample whose mutual correlation between each of said samples and said power spectrum of one cycle of the rotational device is maximum.

7. A rotational-device test system comprising:

a detection means for detecting the time-series data of an amplitude of a drive sound of a rotational device;

a digital conversion means for converting the detected time-series data to digital data;

a power spectrum conversion means for converting a plurality of frame time-series data, obtained by dividing the converted digital time-series data into a plurality of frames at intervals of a predetermined time, to a plurality of power spectra through frequency conversion;

a power detection means for detecting the tone power which is the power of the drive sound contained in a narrow band region, the narrow band region being based on a frequency corresponding to the peak point of the converted power spectrum and having said frequency corresponding to the peak point at its center, and for detecting noise power which is the power of the drive sound in a critical band region width having the peak frequency at its center; and

a dull-sound detection means for detecting whether a predetermined dull sound exists in said drive sound of the rotational device or not, based on the detected tone power and noise power.

8. A rotational-device test method comprising the steps of:

detecting time-series data of an amplitude of a drive sound;

converting the detected time-series data to digital data;

dividing said digital data into a plurality of frames at intervals of a predetermined time and converting a plurality of the divided frame time-series data to a plurality of power spectra through frequency conversion; and

testing a rotational device by classifying said drive sound by comparing the converted power spectrum with the data of the drive sound stored in advance in the form of a power spectrum;

wherein a cycle of said drive sound is specified, and only the data of the drive sound stored in advance which has a frequency substantially identical to the specified cycle of said drive sound is an object of said comparison.

9. The rotational-device test method as set forth in claim 8, wherein said cycle of said drive sound is specified by a first step of determining a candidate point which becomes a candidate for a starting point of said cycle of said drive sound, for each frame, and a second step of specifying said starting point of said cycle from a plurality of determined candidate points.

10. The rotational-device test method as set forth in claim 8, wherein said cycle of said drive sound is specified by a self-correlation.

11. The rotational-device test method as set forth in claim 8, by selecting a maximum point whose self-correlation between the leading frame of said plurality of frames and each of the other frames other than said leading frame is maximum, as a candidate point for determining said one cycle of the drive sound, and by detecting a point whose self-correlation is maximum, the self-correlation having as a starting point the leading time-series data in said leading frame and said candidate point, and based on the frame of said candidate point in the detected point whose self-correlation is maximum and on said leading frame, said cycle of said drive sound is specified.

12. The rotational-device test method as set forth in claim 11, wherein said maximum point is obtained by obtaining a self-correlation having said leading time-series data in said leading frame and leading time-series data in a frame other than said leading frame, as a starting point, for every combination of said leading frame and each of the frames other than said leading frame, and by detecting a point whose self-correlation point obtained is maximum, for every said combination.

13. A rotational-device test method comprising the steps of:

detecting time-series data of an amplitude of a drive sound of a rotational device;

converting the detected time-series data to digital data;

dividing said digital data into frames having a specified length of time and converting a plurality of the frame time-series data into a plurality of power spectra through frequency conversion; and

detecting whether a predetermined dull sound exists in the drive sound of the rotational device or not, based on the converted power spectra.

14. The rotational-device test method as set forth in claim 13, wherein tone power which is the power of the drive sound contained in a narrow band region, which region is based on a frequency corresponding to the peak point of the converted power spectrum and has said frequency corresponding to the peak point at its center, is detected, noise power which is the power of the drive sound in a critical band region width having the peak frequency at its center is detected, and whether a predetermined dull sound exists in said drive sound of the rotational device is detected based on the detected tone power and noise power.

15. The rotational-device test method as set forth in claim 14, wherein a function of more than the second order of a frequency of the drive sound and the energy of the drive sound of said frequency is obtained based on said converted power spectrum, and said peak point is obtained by obtaining a differential coefficient of the obtained function.

16. The rotational-device test method as set forth in claim 13, wherein an average power spectrum is calculated by averaging said converted plurality of power spectra of said frames for every frequency, said tone power and said noise power are detected based on the frequency corresponding to the peak point of the obtained average power spectrum, and whether a predetermined dull sound exists in said drive sound of the rotational device is detected based on the detected tone power and noise power.

17. The rotational-device test method as set forth in claim 16, wherein a function of more than the second order of the frequency of the drive sound and energy of the drive sound of said frequency is obtained based on said average power spectrum, and said peak point is obtained by obtaining a differential coefficient of the obtained function.

18. A rotational-device test method wherein the time-series data of a drive sound of a rotational body is detected, a cycle of said drive sound is specified from said time-series data, and a rotational device is tested by classifying the rotational device by comparing the time-series data of the specified one cycle with the previously stored data of the drive sound having a cycle substantially identical to said specified cycle.