Skip to Main Content
Partial discharge (PD) pattern recognition is widely considered as an effective method to evaluate insulation condition of high voltage apparatuses and distinguish outer pulse interferences. As morphological image tools granulometrics are useful for estimating object sizes and granularity in grayscale image, or characterizing textures based on their pattern spectrum. This paper brings forward a method to extract PD pattern spectrum based on mathematical morphological multi-scale 'open' operation. The pattern spectrum is used as feature of PD grayscale image. Six kinds of typical discharge models are designed and their pattern spectra are calculated, then the two hidden layer neural networks are used for PD pattern recognition. The computation results show that the method proposed in this paper is effective for PD pattern recognition.