Alternative Regression Approach for Data-Driven Power Flow Linearization Methods

Abstract

The knowledge of essential parameters for a power network is necessary for many applications. Evaluating them using traditional methods is computationally intensive and may fail to consider all external factors affecting the system. Therefore, a machine learning-based approach is proposed that predicts these parameters in this paper. The datasets are prepared for IEEE standard test systems and extended for their training and testing. This method can prove helpful to find all the unknown parameters for a power system, especially voltage magnitude and voltage angle, with significantly less error.