Probabilistic load flow evaluation considering correlated input random variables

Abstract: Probabilistic load flow (PLF) is an efficient tool to assess the performance of a power network considering random variables. In this paper, an improved Latin hypercube sampling (LHS) is proposed to solve PLF considering correlated input random variables. The permutation of samples in LHS is treated as a combinatorial optimization problem and handled by a designed genetic algorithm combined with local search (GALS). The developed method is flexible to different measures of dependence and can tackle non-positive definite correlation matrices. Because of the non-normal distributions of output random variables, kernel density estimation (KDE) is used to estimate probability distributions of output data, and different bandwidth selection methods are compared in calculating the bandwidth of KDE. The simulation results of the modified Institute of Electrical and Electronics Engineers (IEEE) 30-bus system and IEEE 118-bus system demonstrate the superiority of the proposed method in solving PLF with dependent random variables.