Detection of disturbances in energy system signals using Gaussian distribution fitness test

In this work, a method based on higher order statistics (HOS) is proposed to detect disturbances in energy system voltage and current waveforms due to faults and various system events. During the normal operation of the system, the noise component imposed on a 50 Hz signal is composed of additive disturbances due to numerous independent events. In this case it is expected that the noise component has a Gaussian distribution. On the other hand, at the instant of a power quality event the noise component would differ and can no longer be considered as Gaussian. In order to detect this difference, the 50 Hz fundamental component of all test signals acquired from the experimental set-up is filtered-out. Consequently, the remaining part of the signal is examined as to whether it can be modelled as Gaussian or not. This is achieved by using skewness and kurtosis values which are derived from the 3^rd and 4^th moments that have small magnitudes for a Gaussian signal. Skewness and kurtosis values are calculated by applying a certain length sliding window on a test data. Later these values are compared with a threshold for testing the data for fitness to a Gaussian distribution and therefore to detect a possible power quality disturbance event.