Generating Capacity Reliability Evaluation Based on Monte Carlo Simulation and Cross-Entropy Methods

This paper presents a new Monte Carlo simulation (MCS) approach based on cross-entropy (CE) method to evaluate generating capacity reliability (GCR) indices. The basic idea is to use an auxiliary importance sampling density function, whose parameters are obtained from an optimization process that minimizes the computational effort of the MCS estimation approach. In order to improve the performance of the CE-based method as applied to the GCR assessment, various aspects are considered: system size, rarity of the failure event, number of different units, unit capacity sizes, and load shape. The IEEE Reliability Test System is used to test the proposed methodology, and also various modifications of this system are created to fully verify the ability of the proposed approach against both, a crude MCS and an extremely efficient analytical technique based on discrete convolution. A configuration of the Brazilian South-Southeastern generating system is also used to demonstrate the capability of the proposed CE-based MCS method in real applications.