Burn-in is a method of eliminating early failures in populations of manufactured items. To burn-in a component or a system means to subject it to a simulated operation for some time (prior to its actual field use). Various optimal burn-in problems have been intensively studied in the literature under the assumption of decreasing or bathtub-shaped failure rates. However, most of these studies have been conducted for homogeneous populations. In this paper, we discuss burn-in for heterogeneous populations and develop approaches that minimize the risks of selecting items with large levels of individual failure rates. Using simple examples, we consider the optimal burn-in time, which minimizes these risks.
|Number of pages||15|
|Journal||Communications in Statistics - Theory and Methods|
|State||Published - 17 Dec 2014|
- Average loss
- Conservative measure
- Heterogeneous population
- Stochastically ordered subpopulations