抄録
In this article, we consider a nonparametric repair-limit replacement (RLR) problem based on the scaled total time on test (TTT) statistics. It should be noted that the estimation procedure for the optimal repair-time limit via the scaled TTT statistics is made on the graph, but is not always stable when the sample size of repair time is small. In order to overcome this problem, we estimate the optimal repair-time limit on an approximated scaled TTT curve. In other words, we apply a smoothing technique based on the natural cubic spline function to interpolate the scaled TTT curve.
