Adaptive sequential importance sampling technique for short-term composite power system adequacy evaluation

The well-known sequential Monte Carlo simulation is a powerful tool to analyse short-term reliability of complicated composite power system. However, the corresponding computational burden is tremendous when applied to a highly reliable system. Aiming at improving the simulation efficiency, an adaptive importance sampling technology is proposed in this study. The proposed method, on the basis of component state duration sampling technique combined with cross-entropy method, is able to provide reliability evaluation of highly reliable non-homogeneous Markov system. The Weibull and lognormal distributions are considered to describe repair time of individual components comprising a system. Through the case studies conducted on a reinforced Roy Billinton reliability test system, it is validated that the proposed method is effective and of high efficiency and the efficiency gain against the crude sequential Monte Carlo simulation is robust to variation of load level and lead timescale. The proposed method is useful and efficient to timely monitor the system operating pressure under different lead timescales.