METHODOLOGIES AND APPLICATION Partial discharge pattern analysis using multi-class support vector machine to estimate cavity size and position in solid insulation B. Vigneshwaran 1 • M. Willjuice Iruthayarajan 1 • R. V. Maheswari 1 Ó Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract Partial discharge (PD) measurement is used for a diagnosis and performance assessment of the solid insulation material inside the high-voltage (HV) equipment. PD measurement indicates the presence of voids, cracks and imperfection present in solid insulation material. The major problem associated with this measured PD signal is heavily contaminated by noise which results in reduction in PD pattern recognition. The objective of this work is to measure and de-noise the PD signal due to cavity and recognize two different size of cavities present in three different locations, namely near HV electrode, center and lower electrode. In first part, the measured PD signal is de-noised using translation invariant wavelet transform. In second part, the three-dimensional (u–q–n) PD patterns are extracted from the de-noised PD data. Then, it is subjected to canny edge detection technique, and the features like horizontal and vertical fractal dimension averages are evaluated using fractal image compression-based semi-variance technique. For classification, multi-class nonlinear sup- port vector machine has been proposed to classify position and size of the cavity based on the PD fingerprints. The findings of this proposed work can be used to design a solid basis for an recognition of cavity size and position in an electrical apparatus. Keywords Partial discharge signal de-noising  Pattern recognition  Support vector machine  Fractal image compression techniques  Three-dimensional (u–q–n) PRPD pattern 1 Introduction In power systems, there is a high demand for diagnosing and condition monitoring of power equipment in an effi- cient and accurate way. PD measurement is acknowledged as a useful diagnostic method with the ability to assess the electrical behavior of insulation material in HV system (Satish and Zaengl 1994). PD clearly localizes the dielec- tric failures in a small region of an insulation system before breakdown occurs in between the two electrodes (Zhang et al. 2019). Under normal operating conditions, electric field stress is distributed uniformly between the electrodes. However, under abnormal conditions like cavities, bubbles or impurities occurring in the HV system cause more stress inside the cavity and leads to insulation degradation. The PD inception voltage mainly depends on the location and size of the cavities present inside the solid insulating materials (Illias et al. 2011, b, 2012; Gutfleisch and Nie- meyer 1995). The major problem in the measurement of PD signal is often hampered by the occurrence of several interferences like white, random and discrete spectral interferences (DSI). This contamination may overwhelm the PD signal and cause errors in assessing the life time of insulation. Therefore, it is necessary to de-noise the acquired PD signals (Tyagi and Vidya 2013). Many researchers proposed several de-noising tech- niques for PD signal in time, frequency and time–fre- quency domains. The time-domain methods are commonly used to minimize periodic pulse interference. But it may fail to retain the amplitude of the original signal. The fre- quency-domain methods were used to de-noise the PD signal but it was limited to the narrow-band noises only (Cunha et al. 2015). Filters and Fourier transform were initially used for de-noising the PD signal. But it could be understood either in time or frequency domain. Once the Communicated by V. Loia. & B. Vigneshwaran bvigneshwar89@gmail.com 1 Department of Electrical and Electronics Engineering, National Engineering College, Kovilpatti, Tamilnadu, India 123 Soft Computing https://doi.org/10.1007/s00500-019-04570-7