Line Selection and Algorithm Selection for Transmission Switching by Machine Learning Methods

Abstract

Since the initial proposal of the Optimal Transmission Switching problem, a mixed integer program and different heuristics have been presented to achieve considerable cost reduction within a practical time frame. This paper proposes two machine learning based methods to further reduce the computation time as well as cutting down the generation cost. The first method is to apply machine learning algorithms to prioritize the possible line switching actions. The second method is to use machine learning to develop effective algorithm selectors among transmission switching algorithms suggested in the literature. The proposed methods are tested on IEEE 118-bus test case and FERC 13867-bus test case. The results demonstrated that both line selection and algorithm selection offer performance benefits over using the single transmission switching algorithm in the previous literature.