Probabilistic Power Flow For AC/VSC-MTDC Hybrid Grids Based On Improved Unscented Transformation Method

With the development of society and technology,the new energy bases will be connected to AC grid based on Voltage Source Control-Multiple Terminal Direct Current system(VSC-MTDC).Future power system will present an operational scenario with complex AC/VSC-MTDC hybrid grids and high proportion of new energy permeation.In order to reveal the impact of fluctuant new energies on AC/DC hybrid grids,it is meaningful to calculate the probabilistic power flow(PPF)for AC/VSC-MTDC hybrid grids.Although the PPF methods for pure AC grid have been developed several years and have reaped rich fruits,there are still lots of challenges of PPF analysis for AC/VSC-MTDC hybrid grids.For example,how to handle the high-dimensional and correlated random variables in AC/DC hybrid grids is a problem.Meanwhile,the random variables may follow arbitrary distributions with linear or non-linear correlation.How to handle it is still challenging.Unscented Transformation(UT)method is an excellent in approximation methods.There are many advantages of PPF analysis by using UT method.This dissertation presents the improved UT methods,according to the features of realistic AC/VSC-MTDC hybrid grids.Firstly,when the traditional UT(TUT)method is used to handle the correlated and high-dimensional random variables following normal distributions,the PPF's accuracy will decrease with the uncertain sources increase.Hence,the Scaled Technology for Unscented Transformation(STUT)method is proposed to deal with the problem.Meanwhile,the calculation strategy of scaled parameter of STUT is proposed.The STUT method not only inherits the advantages of TUT method,but also can flexibly adjust the locations of its samples points,leading to improvement of PPF's accuracy.The effectiveness of STUT is verified by modified IEEE-300 bus system.Secondly,the sample points selected by STUT are hard to approximate the asymmetric distributions.However,in realistic AC/VSC-MTDC hybrid grid,there are a great amount of random variables following asymmetric distributions even with Pearson correlation.Hence,a PPF method based on STUT and NATAF transformation is proposed.With using NATAF transformation,the STUT method is adopted to select the sample points from normal distributions rather than asymmetric distributions,leading to the improvement of PPF's accuracy.The effectiveness of the proposed PPF method is verified by modified IEEE-118 bus system.Meanwhile,the NATAF transformation is generally calculated by Root Searching Method(RSM),which is extremely time-consuming.In this dissertation,the Gauss-Hermite polynomials and Polynomial approximation theory are used to calculate the NATAF transformation(GHM).The complex NATAF transformation can be approximated by a sample polynomial,which can improve the computation efficiency.The effectiveness of the proposed GHM method is verified by several experiments.Thirdly,the STUT method,in general,can deal with the random variables with Pearson correlation.Nevertheless,there is non-linear correlation existing in random variables in AC/VSC-MTDC hybrid grid.Hence,the Copula theory and STUT method are combined to solve the PPF problem.The random variables following arbitrary distributions with rank correlation are transformed into Gaussian distributions,then the STUT is adopted to select sample points from Gaussian distributions.The effectiveness and superiority of the proposed method is verified by modified IEEE-188 bus system.