Differential Protection of ISPST Using Chebyshev Neural Network ‎

Abstract

An Indirect Symmetrical Phase Shift Transformer (ISPST) represents both electrically connected and magnetically coupled circuits, which makes it unique compared to a power transformer. Effective differentiation between transformer inrush current and internal fault current is necessary to avoid incorrect differential relay tripping. This research proposes a system that uses a Chebyshev Neural Network (ChNN) as a core classifier to distinguish such internal faults. For simulations, we used PSCAD/EMTDC software. Internal faults and inrush have been simulated in various ways using various ISPST parameters. A large, simulated dataset is used, and performance is recorded against different sized ISPSTs. We observed an overall accuracy greater than 99%. The ChNN classifier generated exceptionally favorable results even in case of noisy signal, CT saturation, and different ISPST parameters.

Keywords

• energization
• internal fault
• chebyshev neural network (chnn)
• ispst
• pscad/emtdc.‎