Ultra high speed deterministic algorithm for transmission lines disturbance identification based on principal component analysis and Euclidean norm

• Published in: International journal of electrical power & energy systems 2016-09, Vol.80, p.312-324
• Main Authors: Morales, J.A., Orduña, E., Rehtanz, C., Cabral, R.J., Bretas, A.S.
• Format: Article
• Language: English
• Subjects: Algorithms; Deterministic algorithm; Ellipsoidal pattern; Euclidean norm; Lightning; Norms; Principal component analysis; Relay; Strokes; Switching; Transmission lines; Undesirable conditions
• ISSN: 0142-0615; 1879-3517
• DOI: 10.1016/j.ijepes.2016.01.043

•Transient signals detection and identification is presented.•Principal Component Analysis, Ellipsoidal Pattern and Euclidean Norm are used.•The most important parameters of switching operations and lightning strokes are considered.•Different patterns and criteria are determined.•Ultra-high speed novel protection relay algorithms are proposed. Protection devices are designed to provide high sensitivity to transients produced by undesirable conditions like lightning stroke, avoiding their operation under all tolerable events like switching operations. The problem of incorrect operation due to transient phenomena can be handled by two means, one is to allow the transients and provide additional logics in the relay, other means is to damp the oscillation from source side. Protection relays’ not always must trip or send a trip signal and sometimes, only an alarm is necessary. In this context, this research presents a fast and reliable formulation for transmission lines (TLs) switching operations and lightning strokes detection and identification. The proposed methodology is based on Principal Component Analysis (PCA) and Euclidean Norm (EN); by using PCA it is possible to determine that normal operation signals describe a very well defined Ellipsoidal Pattern (EP). In this manner, by calculating the Euclidean Norm (EN) among Principal Components (PCs) for each sample and the origin of the reference Ellipsoidal Pattern, switching operations and lightning strokes are detected and identified. Test results show that the proposed algorithm presents high success on phenomena detection and identification, presenting a high potential for online applications.