Sequential testing of series parallel systems
Gürkan Işık · Sabanci University · 2014
In this thesis, we study the sequential testing problem of 3-level deep Series Parallel systems (SPS). We assess the performance of depth-first permutation (DFP) algorithm that has been proposed in the literature. DFP is optimal for 1-level deep, 2-level deep SPSs and 3-level deep SPSs that consist of identical components. It can be used to test general SPSs. We report the first computational results regarding the performance of DFP for 3-level deep SPSs by comparing its performance with a dynamic version of DFP and a hybrid simulated annealing-tabu search algorithm that we developed. In order to implement the algorithms, we propose an efficient method to compute the expected cost of a permutation strategy. The results of computational experiments for this algorithm and other algorithms proposed in the literature are reported.