Skip to Main Content
In this paper, Latin hypercube sampling (LHS) is investigated in connection with reliability evaluation of a power system. A new sampling technique called discrete Latin hypercube (DLHS) is also proposed. Distributions of reliability indexes resulting from two sampling techniques are presented and analyzed along with those from Monte Carlo (MC) sampling. A comparison among LHS, DLHS and traditional MC for reliability analysis is made. The LHS and DLHS are shown to be more effective than MC for obtaining distributions of indices that are close to the real distributions. The distributions of indices are useful in risk analysis and certain stochastic optimization problems. The test system is a 12-area power system which is based on the data from an actual multi-area system.