A machine learning-based approach for comprehensive fault diagnosis in transmission lines. (July 2022)

• Record Type: Journal Article
• Authors: França, Isternândia Araújo; Vieira, Cynthia Wanick; Ramos, Daniel Correa; Sathler, Lara Hoffmann; Carrano, Eduardo G.
• Is Part Of: Computers & electrical engineering. Volume 101(2022)
• Publication Date: 2022-07
• Subjects: Fault diagnosis -- Fault classification -- Machine learning -- Feature engineering -- Power transmission; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.108107
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: Faults in transmission lines can cause several problems to transmission utilities. An adequate diagnosis of such events can improve service restoration time and, consequently, asset availability. This work proposes a machine learning-based approach to diagnose transmission line faults. The oscillography files of the fault case are processed and applied to four different classifiers, which must identify: phases involved, electrical nature, impedance level, and, fault cause. The method, which is inherently robust regarding signal noise and lack of synchronization between relays, combines Decision Trees, feature engineering, and optimization to perform the fault diagnosis based only on relay measures. It was tested in two fault data sets, one containing 24, 000 faults generated using a Real Time Digital Simulator and another one with 46 real faults. The results obtained were very promising, with accuracy values close to 100% for involved phase and fault nature, 90% for impedance level, and 80% for fault cause. Graphical abstract: Highlights: A ML approach is proposed for comprehensive fault diagnosis in transmission lines. Fault phases, nature, impedance, and cause are inferred from oscillographs. The approach is very robust regarding noise and lack of synchronism between relays. It relies on very few adjustable parameters. Most internal parameters are optimized. Results on RTDS simulated and real fault cases are provided.