A Framework for Analytical Power Flow Solution Using Gaussian Process Learning

This paper proposes a novel analytical solution framework for power flow (PF) solutions in active distribution networks under uncertainty. We use the Gaussian process (GP) regression to learn node voltage as function of effective bus load or negative net-injection vector. The proposed approximation is valid over subspace and provides an understanding system behavior uncertainty via GP interpretability. interpret relative variation extent different voltages using quality ratio (QR) defined based on xmlns:xlink="http://www.w3.org/1999/xlink">hyper-parameters GP. Further, application calculation limit violation probability dominant influencer ranking has also been presented. Through test simulations 33-bus 56-bus systems, method achieves low mean absolute error (MAE) order E-05 (pu) magnitude E-04 (rad) angle. discussions salient features comparative analysis with large-scale Monte-Carlo simulations, state-of-art methods presented applications.