A VALIDATION STUDY OF ESTIMATION METHODS FOR THREE PARAMETERS WEIBULL DISTRIBUTION BASED ON REAL AND HYPOTHETICAL DATA

• Main Author: Dalifa Tafakirey Z. B., Mouna
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/71442
• Institution: Institut Teknologi Bandung

A known used continuous distribution for reliability and life data is a three-parameter Weibull distribution, where the parameters are known as scale parameter (?), the shape parameter (?), and the location parameter (?). In the field of aviation, the three-parameter Weibull becomes a model for estimating the reliability of aircraft maintenance. From existing study [2, 9], the parameters were estimated using methods such as maximum likelihood estimation, Bayesian estimation and the Weibull paper plot (WPP) with Trial-and-Error , though it has yet to be experimented with real data given by Maintenance, Repair and Overhaul (MRO) organization, referred to as MRO A. In addition to real data, the methods will also be experimented with hypothetical data. The parameters obtained will be analyzed and studied among each method and also the robust method will depend on the shape parameter, ? to its corresponding failure mode and also the failure rate graph. When removal occurs in an MRO, the component is examined and checked for the problem. The data obtained includes failed components of an aircraft. MRO A does not have a determined method for analyzing failure data, it can be seen from various models which are being used namely, one and two parameter exponential distribution, two and three Weibull distribution, normal distribution and lognormal distribution. MRO A can benefit from the study of the estimation methods as it can be used for future reliability purposes. To conclude, from this thesis the WPP with trial and error is a more robust method, as the ? value for both real and hypothetical data gave a value similar to the ? of the failure mode and also the failure rate graph.