Comparison on Three Unscented Transformation Methods for Solving Probabilistic Load Flow

In general, the UT methods are regarded as the promising tools to solve the probabilistic load flow calculation problem as the correlated random input variables existing in power system can be tackled inherently, with low computational burden and high computational accuracy. In this paper, three different specific UT methods, including traditional UT method with symmetric sampling strategy (SUT), spherical simplex UT (SSUT) and high-order UT (HOUT), are presented and tested on the modified IEEE 118-bus system, under impact of different number of input random variables and diverse correlation levels amongst some of them. Based on the comparison results, it reveals that SUT is able to achieve a good balance between computational speed and accuracy in most cases. Similar conclusion can be drawn by the HOUT method, but it is implied at the price of a considerably higher computational burden. In addition, SSUT merely behaves well in low-dimensional random variables problem.