WEIBULL PARAMETER ESTIMATION USING MACHINE LEARNING WITH RANDOM FOREST METHOD
The aviation industry requires an effective maintenance program to ensure safety and reduce costs associated with aircraft downtime. Reliability analysis using the Weibull distribution enables the estimation of component failure probabilities from time-to-failure (TTF) data, helping to establish saf...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/86554 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | The aviation industry requires an effective maintenance program to ensure safety and reduce costs associated with aircraft downtime. Reliability analysis using the Weibull distribution enables the estimation of component failure probabilities from time-to-failure (TTF) data, helping to establish safe usage limits. However, traditional methods like Maximum Likelihood Estimation (MLE) and Bayesian Estimation are often inaccurate with limited data. This study proposes a machine learning method, Random Forest, to improve the accuracy and robustness of Weibull parameter estimation, which includes the shape (?), scale (?), and location (?) parameters, in conditions of limited data.
This research compares the performance of the Random Forest model with MLE and Bayesian Estimation across three Weibull parameter scenarios: ? < 1, ? = 1, and ? > 1. The results show that Random Forest provides more accurate and robust estimates for all three Weibull parameters compared to traditional methods. This method has proven to be more consistent in capturing the characteristics of the Weibull distribution, even with limited and randomly selected TTF data. These findings reinforce the use of Random Forest as an alternative method for Weibull parameter estimation with limited data.
|
|---|