Farklı Sıralı Küme Örnekleme Yöntemlerine Dayalı Güvenilirlik Tahmini

Abstract

In the literature, the system reliability R=P(Y<X) estimation, with X power and Y stress, is generally considered based on the simple random sampling (SRS) method. However, recent studies have proven that R estimation based on ranked set sampling (RSS) gives more effective results than standard estimators. In the context of the thesis, first of all, maximum likelihood (MO) estimators based on RSS were obtained. However, when this method is used, the estimators have no analytical solution. For the explicit solution of the estimators, the proposed modified maximum likelihood (MMO) method was used as a non-iterative method. In this thesis, the MO and MMO estimators of R are examined based on the different of the RSS. The modifications considered are moving extreme ranked set sampling (MERSS), L ranked set sampling (LRSS), and median ranked set sampling (MRSS). Finally, the performance comparison of the proposed MO and MMO estimators was done by Monte-Carlo simulation study. In addition, it has been seen that the proposed estimators with the real jute fiber dataset, which is widely used in the literature, are more effective than the standard estimators.