DATA ANALYSIS APPARATUS AND METHOD

• US 20080162081A1
• Filed: 09/20/2007
• Published: 07/03/2008
• Est. Priority Date: 12/27/2006
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Sato, Makoto
• Granted Patent: US 7,472,024 B2
• Time in Patent Office: Days
• US Class Current: 702/184
• CPC Class Codes: G06F 11/008 Reliability or availability analysis

Abstract

There is provided with a data analysis method, including: providing a database storing for each of devices an example that includes inspection time, the degradation level, operation characteristics, and replacement information; providing a evaluation maintenance strategy including an evaluative inspection time, an evaluation time and evaluative replacement condition; providing a prediction model for predicting the degradation level of a device from operation characteristics thereof; estimating by using the database an estimated number of devices that have already been replaced before the evaluation time, evaluating the degradation level at the evaluative inspection time of each of as many devices as estimated number and devices subjected to inspection of same periodic inspection time as the evaluation time; and evaluating the degradation level at the evaluation time of devices whose degradation level at the evaluation inspection time does not satisfy the replacement condition.

Claims

1. A data analysis apparatus for evaluating a maintenance strategy for a plurality of devices, comprising:

a database configured to store first case examples of the devices, each including

inspection time information which indicates an inspection for which of a plurality of periodic inspection times has been performed,

a degradation level of the device,

operation characteristics of the device, and

replacement information indicating whether the device should be replaced or not;

a maintenance strategy storage configured to store a evaluation maintenance strategy including

an evaluative inspection time which indicates a time at which a periodic inspection should be conducted instead of the plurality of periodic inspection times,

an evaluation time indicating a time at which degradation of the devices is to be evaluated and which is later than the evaluative inspection time, in a case that the periodic inspection is conducted at the evaluative inspection time instead of the plurality of periodic inspection times, and

a replacement condition indicating a condition for replacing the device based on the degradation level of the device;

a degradation prediction model storage configured to store a degradation prediction model for predicting the degradation level of the device from the operation characteristics of the device;

a base case example selector configured to select base case examples which are first case examples whose periodic inspection time precedes the evaluation time and having the replacement information indicating that the replacement should be conducted;

a case example generation and selection unit configured to

estimate by using the database an estimated number of devices that have already been replaced before the evaluation time to not exist in the database, obtain the degradation level at the evaluative inspection time of each of as many devices as the estimated number by using at least the former of the base case examples and the degradation prediction model, and generate first virtual case examples including the degradation level at the evaluation time of devices whose obtained degradation level at the evaluative inspection time does not meet the replacement condition based on the base case examples and the degradation prediction model,

calculate the degradation level at the evaluative inspection time of devices whose first case example have same periodic inspection time as the evaluation time from the first case examples of those devices and the degradation prediction model, and select first case examples of devices whose degradation level at the evaluative inspection time does not meet the replacement condition; and

an evaluator configured to evaluate the evaluation maintenance strategy by using generated first virtual case examples and selected first case examples.

2. The apparatus according to claim 1, wherein

the case example generation and selection unit

separates the base case examples into groups each of which has same inspection time information,

estimates for each of the groups the number of devices that have been replaced in an inspection at the same periodic inspection time as that of the group based on a ratio of the number of base case examples included in the group and the number of first case examples that have the same inspection time information as that of the group and have replacement information indicating that the replacement should not be conducted, and

obtains for each of the groups the degradation level at the evaluative inspection time of each of as many devices as estimated number using at least the former of individual base case examples in the group and the degradation prediction model, and generates first virtual case examples including the degradation level at the evaluation time for devices whose degradation level at the evaluative inspection time does not meet the replacement condition based on the base case examples in the group and the degradation prediction model.

3. The apparatus according to claim 2, wherein the degradation prediction model has an individual variability coefficient representing deviation from an average degradation level and the operation characteristics of the device, as parameters,

the case example generation and selection unit

calculates the individual variability coefficient for each base case example in each group by utilizing the degradation prediction model,

when the number of base case examples contained in the each group is less than the number of devices that have been replaced in an inspection at the same periodic inspection time as that of the group, generates as many copies of any base case example contained in the group as a difference in number,

when the periodic inspection time of group is different from the evaluative inspection time,

calculates the individual variability coefficients for the base case examples obtained by copying by use of the individual variability coefficients for copied base case examples in the group and a random number,

estimates operation characteristics at the evaluative inspection time from each of the base case examples and each of base case examples obtained by copying, and

calculates the degradation level at the evaluative inspection time from estimated operation characteristics, the individual variability coefficient, and the degradation prediction model for each of the base case examples and each of base case examples obtained by copying, and

when the periodic inspection time of the group is the same as the evaluative inspection time, obtains a degradation level contained in each of the base case examples and base case examples obtained by copying as the degradation level at the evaluative inspection time.

4. The apparatus according to claim 2, wherein

the degradation prediction model has an individual variability coefficient representing deviation from an average degradation level and the operation characteristics of the device, as parameters,

the case example generation and selection unit

calculates the individual variability coefficient for each base case example in the each group by utilizing the degradation prediction model, and

when the number of base case examples contained in each group is greater than the number of devices that have been replaced in an inspection at the same periodic inspection time as that of the group, selects as many base case examples as the number of devices from each group,

when the periodic inspection time of group is different from the evaluative inspection time, estimates operation characteristics at the evaluative inspection time from each selected base case example, and calculates the degradation level at the evaluative inspection time from estimated operation characteristics, the individual variability coefficient, and the degradation prediction model for each selected base case example, and

when the periodic inspection time of the group is the same as the evaluative inspection time, obtains a degradation level contained in each selected base case example as the degradation level at the evaluative inspection time.

5. The apparatus according to claim 2, wherein

the degradation prediction model has an individual variability coefficient representing deviation from an average degradation level and the operation characteristics of the device, as parameters,

the case example generation and selection unit

calculates the individual variability coefficient for each base case example in each group by utilizing the degradation prediction model,

when the number of base case examples contained in each group is greater than the number of devices that have been replaced in an inspection at the same periodic inspection time as that of the group, selects two or more base case examples for at least one replaced device, and selects for each of remaining replaced devices from base case examples other than the two or more base case examples, and generates a base case example for the at least one device by using the two or more base case examples,

when the periodic inspection time of the group is different from the evaluative inspection time, calculates the individual variability coefficient for the generated base case example by using the degradation prediction model, estimates operation characteristics at the evaluative inspection time from each selected base case example and the generated base case example, and calculates the degradation level at the evaluative inspection time from estimated operation characteristics, the individual variability coefficient, and the degradation prediction model for each of the selected base case examples and the generated base case example, and

when the periodic inspection time of the group is the same as the evaluative inspection time, obtains a degradation level contained in each selected base case example and the generated base case example as the degradation level at the evaluative inspection time.

6. The apparatus according to claim 1, wherein the case example generation and selection unit calculates the operation characteristics of devices which have inspection time information indicating the same periodic inspection time as the evaluation time from the first case examples of the devices, calculates a degradation level from calculated operation characteristics and the degradation prediction model for each device, and selects the first case examples of devices whose degradation level calculated does not meet the replacement condition.

7. The apparatus according to claim 1, wherein the evaluator calculates an average degradation level by using the generated first virtual case examples and the selected first case examples.

8. The apparatus according to claim 1, wherein the evaluator calculates a probability distribution of degradation levels by using the generated first virtual case examples and the selected first case examples, and calculates a probability that the degradation level is a predetermined value or higher from the probability distribution.

9. A data analysis apparatus for evaluating a maintenance strategy for a plurality of devices, comprising:

a database configured to store one or more second case examples of each of the devices, each including

inspection time information which indicates an inspection for which of a plurality of periodic inspection times has been performed,

the degradation level of the device,

operation characteristics of the device, and

replacement information indicating whether the device should be replaced or not;

a maintenance strategy storage configured to store a evaluation maintenance strategy including

an evaluative inspection time which indicates a time at which a periodic inspection should be conducted instead of the plurality of periodic inspection times,

an evaluation time indicating a time at which degradation of the devices is to be evaluated and which is later than the evaluative inspection time, in a case that the periodic inspection is conducted at the evaluative inspection time instead of the plurality of periodic inspection times, and

a replacement condition indicating a condition for replacing the device based on the degradation level of the device;

a degradation prediction model storage configured to store a degradation prediction model for predicting the degradation level of the device from the operation characteristics of the device;

a case example generation and selection unit configured to

obtain the degradation level at the evaluative inspection time for each device whose second case example has a periodic inspection time which precedes the evaluation time and having replacement information indicating that the replacement should be conducted, by using at least the former of the second case example of each device and the degradation prediction model, and generate second virtual case examples each including the degradation level at the evaluation time for the device whose obtained degradation level at the evaluative inspection time does not meet the replacement condition based on the second case example of the device and the degradation prediction model,

calculate the degradation level at the evaluative inspection time for each device that have same periodic inspection time as the evaluation time by using at least former of the second case example of the devices and the degradation prediction model, and select the second case example having same periodic inspection time as the evaluation time from among second case examples of each device whose calculated degradation level at the evaluative inspection time does not meet the replacement condition; and

an evaluator configured to evaluate the evaluation maintenance strategy by using generated second virtual case examples and selected second case examples.

10. The apparatus according to claim 9, wherein the evaluator calculates an average degradation level by using the generated second virtual case examples and the selected second case examples.

11. The apparatus according to claim 10, wherein the evaluator calculates a probability distribution of degradation levels by using the generated second virtual case examples and the selected second case examples, and calculates a probability that the degradation level is a predetermined value or higher from calculated probability distribution.

12. A data analysis method, comprising:

providing a database configured to store for each of a plurality of devices a first case example that includes

inspection time information which indicates an inspection for which of a plurality of periodic inspection times has been performed,

a degradation level of the device,

operation characteristics of the device, and

replacement information indicating whether the device should be replaced or not;

providing a evaluation maintenance strategy including

an evaluative inspection time which indicates a time at which a periodic inspection should be conducted instead of the plurality of periodic inspection times,

an evaluation time indicating a time at which degradation of the devices is to be evaluated and which is later than the evaluative inspection time, in a case that the periodic inspection is conducted at the evaluative inspection time instead of the plurality of periodic inspection times, and

a replacement condition indicating a condition for replacing the device based on the degradation level of the device;

providing a degradation prediction model for predicting the degradation level of the device from the operation characteristics of the device;

selecting base case examples which are first case examples whose periodic inspection time precedes the evaluation time and having the replacement information indicating that the replacement should be conducted;

estimating by using the database an estimated number of devices that have already been replaced before the evaluation time to not exist in the database, obtain the degradation level at the evaluative inspection time of each of as many devices as the estimated number by using at least the former of the base case examples and the degradation prediction model, and generate first virtual case examples including the degradation level at the evaluation time of devices whose obtained degradation level at the evaluative inspection time does not meet the replacement condition based on the base case examples and the degradation prediction model,

calculating the degradation level at the evaluative inspection time of devices whose first case example have same periodic inspection time as the evaluation time from the first case examples of those devices and the degradation prediction model, and select first case examples of devices whose degradation level at the evaluative inspection time does not meet the replacement condition; and

evaluating configured to evaluate the evaluation maintenance strategy by using generated first virtual case examples and selected first case examples.

13. A data analysis method, comprising:

providing a database configured to store for each of a plurality of devices one or more second case examples each including

inspection time information which indicates an inspection for which of a plurality of periodic inspection times has been performed,

the degradation level of the device,

operation characteristics of the device, and

replacement information indicating whether the device should be replaced or not;

providing a evaluation maintenance strategy including

an evaluative inspection time which indicates a time at which a periodic inspection should be conducted instead of the plurality of periodic inspection times,

an evaluation time indicating a time at which degradation of the devices is to be evaluated and which is later than the evaluative inspection time, in a case that the periodic inspection is conducted at the evaluative inspection time instead of the plurality of periodic inspection times, and

a replacement condition indicating a condition for replacing the device based on the degradation level of the device;

providing a degradation prediction model for predicting the degradation level of the device from the operation characteristics of the device;

obtaining the degradation level at the evaluative inspection time for each device whose second case example has a periodic inspection time which precedes the evaluation time and having replacement information indicating that the replacement should be conducted, by using at least the former of the second case example of each device and the degradation prediction model, and generating second virtual case examples each including the degradation level at the evaluation time for the device whose obtained degradation level at the evaluative inspection time does not meet the replacement condition based on the second case example of the device and the degradation prediction model,

calculating the degradation level at the evaluative inspection time for each device that have same periodic inspection time as the evaluation time by using at least former of the second case example of the devices and the degradation prediction model, and selecting the second case example having same periodic inspection time as the evaluation time from among second case examples of each device whose calculated degradation level at the evaluative inspection time does not meet the replacement condition; and

evaluating the evaluation maintenance strategy by using generated second virtual case examples and selected second case examples.