PEMBANDINGAN METODE ESTIMASI TIGA PARAMETER PADA DISTRIBUSI WEIBULL

• Main Author: Adji Fitrian Putri, Yasfiana
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/66413
• Institution: Institut Teknologi Bandung

Weibull distribution is used to estimate the probability of failure of an item or component. The Weibull distribution has an important role in reliability and can be expressed in terms of two or three parameters. Those parameters are shape parameter (????), scale parameter (????), and location parameter (?). Since the location parameter in the two-parameter Weibull distribution is neglected, the three-parameter Weibull distribution will generate a better characterization of the data and also could detect component failure before operation. In this research, a comparison of methods, to the find strengths and weaknesses, will be carried out in estimating parameters of three-parameter Weibull distribution. The methods being compared are the Maximum Likelihood Estimation, Bayesian Estimation and the Weibull Paper Plot (WPP) using a Trial and Error method. The comparison of the methods will be carried out in three stages, i.e., determining the steps in calculating parameters, testing them using sample data sets and execute the calculation in Microsoft Excel, and analyzing the test results. The aspects being analyzed are the complexity, flexibility, and accuracy. The comparison results show that the MLE is the easiest method in terms of complexity, whereas the Bayesian Estimation has the most complex calculations. Parameters estimation using WPP with Trial and Error can be executed for all data types and give reasonable results. Meanwhile, the MLE and Bayesian Estimation methods could not give ? negative results.