PERFORMANCE EVALUATION OF SAMPLING TECHNIQUES IN MONTE CARLO SIMULATION-BASED PROBABILISTIC SMALL SIGNAL STABILITY ANALYSIS

摘要

This paper reports on the performance comparison of simple random sampling (SRS) with Latin hypercube sampling (LHS) techniques in a Monte Carlo based probabilistic small signal stability application. Accuracy and robustness of the sampling methods are tested on Single machine infinite bus (SMIB) and on IEEE 16-machine 68 bus test system. The robustness is based on a significant variance reduction when the experiment is repeated 100 times with different sample sizes. The results show that variance in the result when LHS is employed is much less compared with the same sample size of SRS. Given the results presented in this paper about 100 sample size of LHS is enough to produce a reasonable result for practical purposes in probabilistic small signal stability application.