Self-monitoring storage device using neural networks

• US 6,574,754 B1
• Filed: 02/14/2000
• Issued: 06/03/2003
• Est. Priority Date: 02/14/2000
• Status: Expired due to Fees

• Current Assignee: Western Digital Technologies Incorporated (Western Digital Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Smith, Gordon James
• Primary Examiner(s): Geckil, Mehmet B.
• Application Number:
  US09/503,457
• Time in Patent Office: 1,205 Days
• Field of Search: 709/223, 709/224, 709/325, 706/26, 706/27, 706/21, 706/24, 706/23, 714/26, 714/25, 714/33, 714/39, 714/47, 714/42, 711/111, 711/100
• US Class Current: 714/47.2
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G11B 2220/2516 Hard disks
• CPC Class Codes: G11B 27/36 Monitoring, i.e. supervising the progress of recording or reproducing for digital recording G11B20/00 and s.gr.; for monitoring, testing or measuring of TV recorders of the type covered by H04N5/76 and subgroups, see H04N17/06

Abstract

A digital data storage device such as a rotating magnetic disk drive contains an on-board condition monitoring system, comprising a neural network coupled to multiple inputs derived from measured parameters of disk drive operation. The neural network uses a configurable set of weights to compute one or more quantities representing disk drive condition as a function of the various inputs. The weights are stored in a configuration table, which can be overwritten by a host computer. The drive is sold and installed with one set of weights, based on the then existing knowledge of the disk drive designers, and may be updated in the field as the designers acquire experience data by simply writing the weights to the configuration table of the disk drive, without altering disk drive control code or other disk drive features. Preferably, the disk drive designers include as input to the neural network any parameter which might conceivably be useful, even if the designers initially believe that the parameter has no significance. In this case, the designers can assign the parameter a weight of zero during initial release. If subsequent experience then shows that the parameter has some unexpected significance, the neural network can be corrected simply by changing weighting factors, without altering the control programming code.

Claims

1. A rotating disk drive data storage device, comprising:

a disk drive base;

at least one rotatably mounted disk for recording data on at least one surface of said at least one rotatably mounted disk;

a movable actuator supporting at least one transducer head, said actuator positioning said at least one transducer head to access data on said at least one surface of said at least one rotatably mounted disk; and

a controller for controlling the operation of said disk drive data storage device, said controller including a neural network for monitoring said disk drive data storage device, said neural network accepting a plurality of measurable parameters of said disk drive data storage device as input, and producing at least one monitoring output.

2. The rotating disk drive data storage device of claim 1, wherein said controller comprises a programmable processor and a memory storing a control program for execution on said programmable processor, said neural network being implemented as an artificial neural network simulated by said control program executing on said programmable processor.

3. The rotating disk drive data storage device of claim 1, further comprising:

a dedicated storage area on said at least one surface of said at least one rotatably mounted disk, said dedicated storage area for storing a plurality of connection weights, said connection weights being used by said controller to evaluate said neural network.

4. The rotating disk drive data storage device of claim 3, wherein said controller causes said connection weights in said dedicated storage area to be overwritten with updated connection weights in response to a command received from a host system.

5. The rotating disk drive data storage device of claim 3, wherein said dedicated storage area is a SCSI mode page.

6. The rotating disk drive data storage device of claim 1, wherein said plurality of measurable parameters includes at least one parameter which is measured separately for each said disk surface, the separate measurements of said parameter for each disk surface constituting separate inputs to said neural network.

7. The rotating disk drive data storage device of claim 1, wherein said plurality of measurable parameters includes at least one measure of soft error rate.

8. The rotating disk drive data storage device of claim 1, wherein said plurality of measurable parameters includes at least one measure of flyheight.

9. A method for operating a digital data storage device comprising the steps of:

accessing data stored on said data storage device responsive to requests for data;

periodically triggering a self-evaluation of said digital data storage device;

inputting a plurality of measured parameters of said digital data storage device to a neural network resident in said digital data storage device, responsive to said step of triggering a self-evaluation of said digital data storage device; and

evaluating said neural network in said digital data storage device to produce a self-evaluation of said digital data storage device.

10. The method of claim 9, wherein said neural network is implemented as an artificial neural network simulated by a control program executing on a programmable processor of said digital data storage device.

11. The method of claim 9, wherein digital data storage device is a rotating magnetic disk drive storage device.

12. The method of claim 9, further comprising the steps of:

retrieving a plurality of neural network connection weights from a dedicated read/write data storage area of said data storage device; and

using said plurality of neural network connections weights to evaluate said neural network.

13. The method of claim 12, further comprising the step of:

overwriting said neural network connection weights in said dedicated storage area with updated neural network connection weights in response to a command received from a host system.

14. A program product for controlling the operation of a digital data storage device, said program product comprising:

a plurality of processor executable instructions recorded on signal-bearing media, wherein said instructions, when executed by at least one programmable processor, cause the storage device to perform the steps of;

accessing data stored on said data storage device responsive to requests for data;

periodically triggering a self-evaluation of said digital data storage device;

inputting a plurality of measured parameters of said digital data storage device to an artificial neural network simulated by said program product executing on said programmable processor, responsive to said step of triggering a self-evaluation of said digital data storage device; and

evaluating said artificial neural network in said programmable processor to produce a self-evaluation of said digital data storage device.

15. The method of claim 14, wherein digital data storage device is a rotating magnetic disk drive storage device.

16. The method of claim 14, further comprising the steps of:

retrieving a plurality of neural network connection weights from a dedicated read/write data storage area of said data storage device; and

using said plurality of neural network connections weights to evaluate said neural network.

17. The method of claim 16, further comprising the step of:

overwriting said neural network connection weights in said dedicated storage area with updated neural network connection weights in response to a command received from a host system.

18. A method for monitoring a plurality of storage devices, said plurality being devices of the same model of storage device, said method comprising the steps of:

designing said model of storage device having a neural network for self monitoring, said neural network accepting a plurality of measurable parameters as input, and at least one monitoring variable as output;

assigning a first set of weights to connections of said neural network in a plurality of disk drive storage devices of said model of storage device;

obtaining data representing performance of said model of storage device;

generating a second set of weights to connections of said neural network, said second set being generated using said data representing performance of said model of storage device, said second set of weights being different from said first set of weights; and

replacing said first set of weights with said second set of weights in a plurality of storage devices of said model of storage devices.

19. The method of claim 18, wherein said neural network is implemented as an artificial neural network simulated by a control program executing on a programmable processor of said storage device.

20. The method of claim 18, wherein said step of replacing said first set of weights with said second set of weights comprises overwriting said first set of weights with said second set of weights, said weights being written to a common dedicated storage area of each storage device of said model of storage device.

21. The method of claim 20, wherein said weights are written to a SCSI mode page reserved location.