Studies on variance reduction technique of Monte Carlo simulation in composite system reliability evaluation

A new variance reduction technique of Monte Carlo simulation - fission and roulette method is presented in the paper. The proposed approach reduces the variance of simulation and speeds up the computation dramatically. Comparing with other variance reduction techniques, this method can adjust sampling scheme according to concrete conditions of the given systems. For the states that may have load curtailment, we can increase the sampling number and for the states that may have no load curtailment, we can decrease the sampling number. Thus, the sampling scheme grasps the key points and improves the convergency performance. Furthermore, the method includes the judgment of the system characteristic, so it possesses a good adaptability to different systems. The calculation results of IEEE-RTS show that our method reduces the computational effort evidently.