Estimating Density Functions for Probabilistic Power Flow Using Invertible Neural Networks
This work proposes an Invertible Mixed Neural Flow (IMNF) framework to accurately estimate voltage probability density functions in Probabilistic Power Flow (PPF). It uses invertible neural networks with a change-of-variables theorem to map power injections to voltages, achieving state-of-the-art performance as both a PF surrogate and a density estimator. The method is efficient, requiring fewer samples than traditional methods.