Probabilistic load flow evaluation with hybrid Latin Hypercube Sampling and multiple linear regression

Probabilistic load flow (PLF) has gained wide attention in power system planning and operation as an efficient tool to analyze the influences of random variables. In this paper, Latin Hypercube Sampling combined with multiple linear regression (LHS-MLR) is proposed to solve PLF evaluation considering correlated variables. Based on the relationship between linear correlation and linear regression, an algorithm is designed to arrange the orders of samples to introduce desired dependences among variables. The proposed method is compared with NORTA and Genetic Algorithm using IEEE 9-bus and IEEE 118-bus systems. Simulation results indicate that LHS-MLR is a promising method in PLF evaluation.