Probabilistic Locational Marginal Price by Monte Carlo Simulation with Latin Hypercube Sampling and Scenario Reduction Techniques

Abstract

Leveraging Monte Carlo Simulation (MCS) combined with simple random sampling (SRS) to evaluate probabilistic locational marginal price (P-LMP) requires long computation time and large computer storage. The paper proposes the joint usage of Latin hypercube sampling (LHS) with sample reduction techniques called the fast forward selection (FFS) algorithm into Monte Carlo simulation for calculation of the P-LMP. This fast forward selection algorithm is needed to cut down the number of samples while keeping most of the stochastic information embedded in such samples. The LHS-FFS-based P-LMP is investigated using IEEE 6-bus and 24-bus systems. This method is compared with SRS and LHS only. The LHS-FFS approach is found to be efficient and flexible; therefore, it has the potential to be applied in many power system probabilistic problems.