Ultra-sensitive disturbances detection on transmission lines using principal component analysis

It is known that some transient events are too difficult to be detected immediately; especially high impedance faults, faults that occur at zero-crossing voltage and faults in the relay proximity. A novel approach for detecting transient events and wave distortion, based on Principal Component Analysis is presented. The proposed algorithm is easy to implement and is not time consuming. It only requires computing the dot product of two vectors. The first vector is a discrete signal of the last cycle wave and the second vector is a high-pass filter which reduces the fundamental component of the analyzed signal to zero and only measures deviations of this signal with respect to perfect sinusoidal waves. Therefore it allows for determining how distorted the analyzed wave is.