Failure prediction techniques based on Weibull model
The time for the occurrence of failure in a machine has been predicted using a Weibull model. The model uses the information of past failures and fits it into a probability distribution that yields a prediction of future failures. The operational data used for analysis is a seri...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Theses and Dissertations |
| Language: | English |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/59941 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | The time for the occurrence of failure in a machine has been predicted using a Weibull
model. The model uses the information of past failures and fits it into a probability
distribution that yields a prediction of future failures. The operational data used for
analysis is a series of failure times procured from an industrial machine used in a
manufacturing system. This thesis discusses three methods of parametric estimation of
the Weibull distribution, namely the maximum likelihood estimation, the method of
moments, and the least squares method, and compares their errors in estimation and
develops a graphical approach to help choose the right method for the right application.
In addition, for the maximum likelihood estimation method, we modify the data set into
an interval censored set of data and estimate the parameters for various observation
lengths. The error for various observation lengths has been plotted and a tradeoff is
developed between inspection load and error. This helps to choose an optimal value of
the observation length. Finally, a time-to-failure prediction based on the estimated
parameters is done. |
|---|