Measurement and recognition of Single and multiple partial Discharge pattern using artificial Intelligence techniques

Abstract

In power systems, there is a high demand for diagnosing and Condition Monitoring (CM) of power equipment in an efficient and accurate way. Measurement and recognition of Partial Discharge (PD) is an efficient method for CM of High Voltage (HV) equipments. Analysis of PD activity and its behavior is usually used as an indicator of insulation degradation over time. The major problem associated in this measured PD signal is that it is heavily contaminated by noise which reduces effectiveness of PD pattern ecognition methods. Several researchers proposed different de-noising schemes for PD signals. The proposed work presents recognition of single and multiple PD sources using the characteristics of PRPD patterns of different media. The proposed work is focused on preprocessing techniques employed for PRPD pattern images before feature extraction. The preprocessing includes edge detection, de-noising the PD signal, dimensionality reduction and Affine Transformation (AT). In the first part of the proposed work, an attempt has been made to ecognize the single and multiple PD sources having moderate datasets of PD signature patterns. The first part is further classified into two sections. The first section is to classify the multiple PD sources using a combined algorithm of different edge detection methods along with a box-counting fractal image compression technique. The second section is to recognize two different sizes of a cavity present in three different locations namely near the HV electrode, center and the lower electrode. The measured PD signal is de-noised using ranslation Invariant Wavelet Transform (TIWT) and 3-D(and#966;-q-n) PD patterns are extracted from the de-noised PD data newline