An efficient hybrid technique for correlated probabilistic load flow study with photovoltaic generations

Abstract

This paper proposes an efficient hybrid technique for probabilistic load flow study. A mixture of correlated Gaussian and non-Gaussian as well as discrete distributions is considered for input random variables. Distributions of desired random variables pertaining to the input random variables are found to be multimodal. Analysis using Gaussian mixture approximation is promising in this context, but computational burden increases significantly with the increase in number of discrete random variables. In contrast, the proposed method precisely obtains distribution of desired random variables in considerably less time without compromising accuracy. Multiple input correlations are effectively incorporated. Accuracy of the proposed method is examined in IEEE 14-bus and 57-bus test systems. Results are compared with combined cumulant-Gaussian mixture approximation method and Monte-Carlo simulation. � 2016 IEEE.