Parameterized Linear Power Flow for High Fidelity Voltage Solutions in Distribution Systems

TL;DR

This paper presents a novel parameterized linear power flow model that uses learning-based techniques to achieve high-fidelity voltage solutions in distribution systems across various operating conditions.

Contribution

It introduces a system-specific, self-adjusting parameterization method that enhances the accuracy of linear power flow models without extensive recomputation.

Findings

Improved voltage solution accuracy over simplified models

Effective across multiple loading levels

Validated on six test systems

Abstract

This paper introduces a new model for highly accurate distribution voltage solutions, coined as a parameterized linear power flow model. The proffered model is grounded on a physical model of linear power flow equations, and uses learning-aided parameterization to increase the fidelity of voltage solutions over a wide range of operating points. To this end, the closed-form analytic solution of the parameterization approach is obtained via a Gaussian Process using a deliberately small input sample and without the need for recomputation. The resulting "self-adjusting" parameter is system-specific and controls how accurate the proposed power flow equations are according to loading conditions. Under a certain value of the resulting parameter, the proposed model can fully recover the linearized formulation of a specialized branch flow model for radial distribution systems, the so-called simplified DistFlow model. Numerical examples are provided to illustrate the effectiveness of the proposed model as well as the improvement in solution accuracy for voltage magnitudes over the simplified DistFlow model and several other linear power flow models, at multiple loading levels. Simulations were carried out on six small- and medium-sized test systems.