Automatic threshold of standard deviation to reject noise in raw data of partial discharges

Abstract

The raw data PD measurement results by using a computer usually contain noise that can lead to miss interpretation. So as to obtain an accurate diagnosis, then this noise must be separated from the raw data. In this study, the electrodes used were needle-plane electrode and the samples tested are polymer films placed on the electrode plane and the air gap is located between the needle and polymer films. There are two types of arrangement of the electrodes, the first electrode needle wrapped with the film and the second polymer are not wrapped with polymer films. To separate the data PD of raw-data is necessary algorithms that can automatically select the threshold value of the standard deviation value. To obtain the optimum value, the threshold value selected based on the standard deviation of each value segment. The results show that this method has been used to distinguish between the electrode pattern PD wrapped or not wrapped polymer.