Methods and systems for providing unanticipated demand predictions for maintenance

• US 7,945,427 B2
• Filed: 04/18/2008
• Issued: 05/17/2011
• Est. Priority Date: 04/18/2008
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: Kolasa, Stan A., Guenther, Nicholas A., Nash, April Y.
• Primary Examiner(s): Le; John H
• Application Number:
  US12/105,945
• Publication Number:
  US 20090265118A1
• Time in Patent Office: 1,124 Days
• Field of Search: 702/34, 702/35, 702/81, 702/82, 702/83, 702/181, 702/184, 702/185, 701/3, 701/29, 340/506, 705/9, 714/37, 714/47, 716/4
• US Class Current: 702/184
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

A method for predicting a probability of failure for a component of a platform at a specified time is described. The method includes collecting historical maintenance data relating to the component, selecting, utilizing the collected historical maintenance data, a lifetime distribution model that best fits the historical maintenance data, estimating upcoming component failures using the selected lifetime distribution model, and applying maintenance schedule dates for the platform to the upcoming component failures to determine a likelihood of failure of the component on one of the scheduled maintenance dates.

Claims

1. A computer-based method for predicting a probability of failure for a component of a platform at a specified time, said method comprising:

collecting, within a computer memory, historical maintenance data relating to the component;

selecting, utilizing the collected historical maintenance data, a lifetime distribution model that best fits the historical maintenance data;

estimating, with a computer, upcoming component failures using the selected lifetime distribution model; and

applying maintenance schedule dates for the platform to the upcoming component failures, with the computer, to determine a likelihood of failure of the component on one of the scheduled maintenance dates;

wherein selecting a lifetime distribution model that best fits the historical maintenance data comprises using at least one of a predetermined goodness of fit factor and a lack of fit test, to determine a fit of the collected historical maintenance data to each of the available lifetime distribution models; and

wherein using the at least one of the predetermined goodness of fit factor and the lack of fit test comprises evaluating a fit of the historical maintenance data with a lifetime distribution model using a test statistic that describes how closely resulting data emulates each available lifetime distribution model.

2. A computer-based method according to claim 1 wherein selecting a lifetime distribution model comprises selecting one of a normal distribution model, an exponential distribution model, a Weibull distribution model, a lognormal distribution model, and a gamma distribution model.

3. A computer-based method according to claim 1 wherein selecting a lifetime distribution model comprises contrasting actual failure and lifetime data for the component to several lifetime distribution models.

4. A computer-based method for predicting a probability of failure for a component of a platform at a specified time, said method comprising:

collecting, within a computer memory, historical maintenance data relating to the component;

selecting, utilizing the collected historical maintenance data, a lifetime distribution model that best fits the historical maintenance data;

estimating, with a computer, upcoming component failures using the selected lifetime distribution model;

applying maintenance schedule dates for the platform to the upcoming component failures, with the computer, to determine a likelihood of failure of the component on one of the scheduled maintenance dates;

comparing a distribution of residuals with a distribution of errors in the selected lifetime distribution model using a quartile plot; and

interpreting abnormal deviations from the error distribution as a potential deficiency in the selection of a particular lifetime distribution model.

5. A computer-based method for predicting a probability of failure for a component of a platform at a specified time, said method comprising:

collecting, within a computer memory, historical maintenance data relating to the component;

selecting, utilizing the collected historical maintenance data, a lifetime distribution model that best fits the historical maintenance data;

estimating, with a computer, upcoming component failures using the selected lifetime distribution model;

applying maintenance schedule dates for the platform to the upcoming component failures, with the computer, to determine a likelihood of failure of the component on one of the scheduled maintenance dates;

creating a cumulative distribution function for the selected lifetime distribution model for the component to establish a compounding aggregation of failure throughout its lifetime;

generating a hazard rate distribution, illustrating a rate at which the component fails, for the elected lifetime distribution model;

wherein applying maintenance schedule dates for the platform to the upcoming component failures comprises;

applying the scheduled maintenance dates to the lifetime distribution model by extracting and utilizing the cumulative distribution function and hazard rate distribution to evaluate the likelihood of component failure on the actual scheduled maintenance date;

representing the likelihood of component failure as a failure propensity coefficient; and

presenting the failure propensity coefficient to a maintenance scheduler.

6. A system for predicting a probability of failure for a component of a platform at a specified time, said system including at least one computer programmed to cause a computer processor to:

receive and store historical maintenance data relating to the component;

select a lifetime distribution model that best fits the stored historical maintenance data;

estimate an upcoming component failure using the selected lifetime distribution model; and

apply maintenance schedule dates for the platform to the upcoming component failure estimate to determine a likelihood of failure of the component on one of the scheduled maintenance dates;

wherein to select a lifetime distribution model that best fits the stored historical maintenance data, said computer is programmed to use a predetermined goodness of fit factor to determine a fit of the stored historical maintenance data to each of the available lifetime distribution models; and

wherein said computer is programmed to evaluate a fit of the stored historical maintenance data with the lifetime distribution model using a test statistic that describes how closely resulting data emulates each available lifetime distribution model.

7. A system according to claim 6 wherein said computer is programmed with at least one of a normal distribution model, an exponential distribution model, a Weibull distribution model, a lognormal distribution model, and a gamma distribution model.

8. A system according to claim 6 wherein said computer is programmed to contrast actual failure and lifetime data for the component against a plurality of lifetime distribution models.

9. A system according to claim 6 wherein to estimate upcoming component failures, said computer is programmed to:

create a cumulative distribution function for the selected lifetime distribution model for the component to establish a compounding aggregation of failure throughout its lifetime; and

generate a hazard rate distribution that illustrates a rate at which the component fails, for the elected lifetime distribution model.

10. A system according to claim 9 wherein to apply maintenance schedule dates for the platform to the upcoming component failures, said computer is programmed to:

apply the scheduled maintenance dates to the lifetime distribution model by extracting and utilizing the cumulative distribution function and hazard rate distribution to evaluate the likelihood of component failure on the actual scheduled maintenance date;

represent the likelihood of component failure as a failure propensity coefficient; and

present the failure propensity coefficient to a maintenance scheduler.

11. A system for predicting a probability of failure for a component of a platform at a specified time, said system including at least one computer programmed to cause a computer processor to:

receive and store historical maintenance data relating to the component;

select a lifetime distribution model that best fits the stored historical maintenance data;

estimate an upcoming component failure using the selected lifetime distribution model; and

apply maintenance schedule dates for the platform to the upcoming component failure estimate to determine a likelihood of failure of the component on one of the scheduled maintenance dates;

wherein said computer is programmed to;

compare a distribution of residuals with a distribution of errors in the selected lifetime distribution model using a quartile plot; and

interpret abnormal deviations from the error distribution as a potential deficiency in the selection of a particular lifetime distribution model.

12. A computer-based maintenance and repair method associated with low volume and long lead time parts for a platform, said method comprising:

utilizing a depot maintenance schedule within a computer memory to identify scheduled visits of a platform to a depot level maintenance area;

retrieving, at a computer associated with the computer memory, maintenance data relating to low volume and long lead time parts associated with the platform;

mining the retrieved maintenance data, with the computer, to determine a lifetime distribution model that best fits historical part failures;

applying, using the computer, a dynamic model to each part lifetime to produce a failure propensity coefficient for each part with respect to each scheduled depot level maintenance area visit; and

building an estimated demand plan for upcoming part failures, based on the failure propensity coefficients and the scheduled depot level maintenance area visits.

13. A maintenance and repair method according to claim 12 further comprising commencing procurement of at least one low volume and long lead time replacement part if the failure propensity coefficient indicates a confidence level that the part will fail at a scheduled depot level maintenance area visit.

14. A maintenance and repair method according to claim 13 further comprising:

verifying an operational status of at least one low volume and long lead time replacement part during a depot level maintenance area visit for the platform on which the part is deployed; and

upon determining the part has not failed, and does not need to be replaced, updating the lifetime distribution model for the part; and

re-verifying the selected lifetime distribution model provides the best fit for part failures.

15. A maintenance and repair method according to claim 13 further comprising:

verifying an operational status of at least one low volume and long lead time replacement part during a depot level maintenance area visit for the platform on which the part is deployed; and

upon determining the part has not failed, and does not need to be replaced, re-evaluating other upcoming depot level maintenance area visit determine the next best utilization for the unused, but now procured low volume and long lead time replacement part.