Research on state evaluation and risk assessment for relay protection system based on machine learning algorithm

Liming Ying,Yongtian Jia,Wenan Li
DOI: https://doi.org/10.1049/iet-gtd.2018.6552
2020-08-04
Abstract:The relay protection system plays an important role in ensuring the stable operation of power systems. Combined with operation data collected from a region in China, this study is aimed at providing a reliable quantitative basis for relay protection systems' operating maintenance by the aid of a semi-supervised Mahalanobis distance machine learning algorithm. The evaluation result is first applied as a training set on the basis of the analytic hierarchy process fuzzy synthetic evaluation. Then, contrastive analysis is conducted in terms of accuracy, processing time, and feasibility. It includes comparative cases with a supervised multiple regression analysis algorithm and unsupervised K-means algorithm. The comparison result reveals that the algorithm can effectively and accurately predict the running state of the equipment and offer a quantitative reference for relative maintenance strategy.
What problem does this paper attempt to address?