论文信息 - On Lifetime Optimization of Boolean Parallel Systems with Erlang Repair Distributions

On Lifetime Optimization of Boolean Parallel Systems with Erlang Repair Distributions

We assume a Boolean parallel system with exponentially distributed component failure times. In order to maximize the lifetime of the system we consider a repairman with Erlang-k distributed repair time. By extending the classical exponential case k = 1 to k ≥ 2 different repair semantics arise in this context. In the case of restart repair semantics we show that a repairman should have few Erlang phases in order to maximize mean time to failure (MTTF). In the case k ≥ with full memory semantics an optimal repairman policy shows counter-intuitive repair behaviour dependent on the mean repair time.

Markus Siegle | Alexander Gouberman | M. Siegle | A. Gouberman

[1] Michael N. Katehakis,et al. Dynamic Repair Allocation for a k −Out−of− n System Maintained by Distinguishable Repairmen , 1988 .

[2] Roger M. Kieckhafer,et al. On the sensitivity of NMR unreliability to non-exponential repair distributions , 2000, Proceedings. Fifth IEEE International Symposium on High Assurance Systems Engineering (HASE 2000).

[3] Donald R. Smith,et al. Optimal Repair of a Series System , 1978, Oper. Res..

[4] U. Rieder,et al. Markov Decision Processes , 2010 .

[5] Michael N. Katehakis,et al. Optimal Repair Allocation in a Series System , 1984, Math. Oper. Res..

[6] Martin L. Puterman,et al. Markov Decision Processes: Discrete Stochastic Dynamic Programming , 1994 .

[7] Bent Natvig,et al. New Results on the Barlow–Proschan and Natvig Measures of Component Importance in Nonrepairable and Repairable Systems , 2009 .