Investigation of Computational Advantage of using Importance sampling in Monte Carlo Simulation

This paper presents a sensitivity analysis of the composite system reliability evaluation using a cross-entropy based importance sampling method to improve computational efficiency of sequential Monte Carlo simulation (CE MCS). The sensitivity analysis studies how the computational performance of the method and reliability indices are affected by varying the system parameters like peak load and forced outage rates of the generators. The relationship between computation time, simulation parameters, coefficient of variance and number of system cores is also explored. The sequential Monte Carlo simulation is implemented using parallel computing techniques which reduces the computation time. A comparison study is carried out using Simple Monte Carlo method. These methods are tested on an IEEE RTS 79 test system.