Proposal of a bootstrapping methodology to calculate reliability indexes

Abstract

This study proposes an algorithm to simulate population samples, supported by the Monte Carlo and Bootstrap methods, where the sample size is not representative of the universe being studied. This scenario is characteristic of investigation processes or events which span a long time period. This study developed an example with real-life data collected from a harvesting machine, calculating the lifetime reliability gamma index for a hydraulic element. The implementation of the algorithm was developed with the mathematical assistant MATLAB and the block diagram is shown at work. From the analysis of the results it is conclusive that, when the sample size being studied is very small, the proposed methodology is appropriate to estimate the necessary probabilistic distribution and therefore allows to estimate the confidence interval of the reliability index that is being sought.