Multivariate Principal Component Analysis Wavelet for Quick Detection and Isolation of Abnormalities in a Coordinated Protection Scheme

Abstract

Owing to the capital-intensive nature of the power system components and safety
of lives, fast response to intolerable conditions of power system networks becomes
a growing concern for experts in power system monitoring and protection. To
synthesize the noise nature of the signals, the multivariate principal component
analysis (MPCA) wavelet method was used to extract and simplify the fault signals.
An exploit of Daubechies wavelet (db3) was made and decomposed up to level 7.
To have a crack at the composite noise nature, an attempt was made on leveldependent
noise size estimation which was down-sampled by two at each
succeeding level of 7. The results obtained from details, approximations, and
simplified signals show that by far, multi-scale principal component analysis
(MPCA) is better than multi-resolution signal analysis (MRSA) wavelet for signal
detection and simplification.