Abstract
One important issue in reliability and quality engineering is to assess a system reliability properly. Systems complexity increases the costs of miscalculation of reliability. Thus, it is very important to measure reliability accurately. Different classical and robust methods of estimating a system reliability are proposed by different authors. Each may be used under certain conditions. In this research, a robust estimator of a system reliability in its useful life period is proposed. The suggested method not only introduces an estimator robust to the presence of outliers but also performs as good as a classical estimator when there is no outlier. The proposed estimator is a revised M-scale estimator. Evaluation of the suggested estimator by means of simulations and its comparison to other robust and classical estimators of system reliability, applying the efficiency and robustness criteria indicate that it performs much better than the existing estimators.
Acknowledgements
The authors would like to thank the anonymous reviewers and the editor whose comments and suggestions improved greatly the presentation of the paper. Our thanks also to Dr. Orod Ahmadi for his generous help.