Short-term reliability evaluation using control variable based dagger sampling method

Most of the reliability theory and applications are in the planning domain. When reliability concepts are applied in operations or operational planning, computational burden of the algorithms becomes a bottleneck as the speed of execution of algorithms becomes critical. This paper presents an efficient control variable based dagger sampling technique for reducing the computational effort in Monte-Carlo reliability evaluation for composite systems. The proposed variance reduction method is unique in combining the offline calculations with online computation. A large number of system states are simulated to calculate their consequences offline and stored. Short-term reliability sampling uses the information on offline computed consequences to construct a variance reduction function. In short-term reliability evaluation, the outage probability of each component is much smaller compared to long-term system reliability calculations. Dagger sampling is used to reduce the computation time under this condition. Test results on RTS show the improvement and effectiveness in convergence speed.