Reliability Analysis in the Software R

Abstract

The reliability study can be used to analyze the times of faults of the equipment and to determine which distribution is adjusted better to the data. There is software that executes this type of procedure, however, most of them are developed for application in the industrial sector, being generally paid and closed. This work aims to develop a code in R software capable of analyzing and determine which distribution is adjusted better to the data, using goodness of fit tests (numerics and graphics). Famous distributions such as Weibull and Lognormal were implemented, as well as complex distributions, such as the Generalized Gamma. For analysis, the code displays results from various tests such as Likelihood Ratio Test (LRT), Kolmogorov-Smirnov (KS) and Chi-square (χ2), in addition to generating graphics of density, accumulated density, reliability, risk and showing the roles of the probability of distributions. All the code was developed in R, as it is a free platform, so it facilitates the work of researchers and companies in the reliability sector. Three sets of equipment failure time data were analyzed, the results found has been coherent and some superior cases when compared with other works and software.