Probabilistic load flow computation with polynomial normal transformation and Latin hypercube sampling

Abstract

A probabilistic load flow method based on polynomial normal transformation (PNT) and Latin hypercube sampling (LHS) is proposed. The correlation between input random variables has been taken into consideration. The proposed method uses the statistical moments and correlation matrix of input random variables instead of their marginal distribution functions and joint distribution functions, which are very difficult to be obtained, to establish their probability distribution models by PNT and LHS. The statistical moments and probability distribution functions of node voltage and line flow are calculated by Monte Carlo simulation method. Performance of the proposed method is investigated using IEEE 14‐bus and IEEE 118‐bus test systems. The impacts of correlation factor on the statistical moments of power injections and system operation are analysed. Finally, conclusions are duly drawn.