Acoustic Partial Discharge signal analysis using digital signal processing techniques

An important aspect of condition monitoring is the assessment of Partial Discharge as the major cause of High Voltage equipment ageing and degradation. Several methods such as chemical, electric and acoustic are used for PD detection. This paper addresses the practical method for Acoustic PD detection in high voltage Transformer insulation oil and its processing. Digital Signal Processing technique has been adopted using Fast Fourier Transform, Short Time Fourier Transform and Wavelets. Pulse entropy and energy have been computed to understand the pulse behavior at various decomposition levels. A comparative study of Fractal features has been done for a set of pulses obtained from the same measurement system. The diverse behavior of these fractals has been correlated to the internal properties of the insulation medium. The results also prove the validity of certain fractals for classification problems.