Fractional Moments Based Adaptive Scaled Unscented Transformation for Probabilistic Power Flow of AC-DC Hybrid Grids

Probabilistic power flow (PPF) is the fundamental to reveal the influence of stochastic sources on the AC-DC hybrid grids. In the operational PPF analysis, the accurate probability density functions (PDFs) of PPF responses must be obtained in a very short period, which is quite challenging for existing methods. In this paper, a fractional moments based adaptive scaled unscented transformation (ASUT) is proposed to overcome the operational PPF challenge. The ASUT creates a new path to fully catch the probability information, which adaptively selects the sample point sets from each random input variable through the fractional moment assessment. A handful of fractional moments can contain the information identical to the one generated from a great many integer moments (e.g., central or raw moments). This feature is beneficial to present and propagate the abundant probability information using only a few sample points, which enables the reconstruction the PDFs of PPF results accurately by the maximum entropy, leading to the great enhancement in the execution accuracy and efficiency of PPF calculations. Test results on the IEEE 118 bus system integrated with DC systems and a provincial AC-DC hybrid grid in South China validate the effectiveness and advantages of proposed method herein.