Abstract
Partial Discharge (PD) pattern recognition plays an important part in electrical equipment fault diagnosis and maintenance. Feature extraction could greatly affect recognition results. Traditional PD feature extraction methods suffer from high-dimension calculation and signal attenuation. In this study, a novel feature extraction method based on Ensemble Empirical Mode Decomposition (EEMD) and Sample Entropy (SamEn) is proposed. In order to reduce the influence of noise, a wavelet method is applied to PD de-noising. Noise Rejection Ratio (NRR) and Mean Square Error (MSE) are adopted as the de-noising indexes. With EEMD, the de-noised signal is decomposed into a finite number of Intrinsic Mode Functions (IMFs). The IMFs, which contain the dominant information of PD, are selected using a correlation coefficient method. From that, the SamEn of selected IMFs are extracted as PD features. Finally, a Relevance Vector Machine (RVM) is utilized for pattern recognition using the features extracted. Experimental results demonstrate that the proposed method combines excellent properties of both EEMD and SamEn. The recognition results are encouraging with satisfactory accuracy.
Highlights
Partial discharge (PD) detection plays an important role in the evaluation of insulation condition [1]
Partial Discharge fault recognition plays an important part in the insulation diagnosis of electrical equipment
Based on the Intrinsic Mode Functions (IMFs) of Ensemble Empirical Mode Decomposition (EEMD), Sample Entropy is calculated, which is sensitive to the properties of Partial Discharge (PD) signals
Summary
Partial discharge (PD) detection plays an important role in the evaluation of insulation condition [1]. Different PD types may cause diverse damages to equipment insulation [2]. It is meaningful to be able to distinguish between different PD types for electrical equipment repair and maintenance [3,4]. Feature extraction is of great importance during PD pattern recognition. It directly affects the recognition results [5,6,7,8,9]. Chu et al employed statistical distribution parameters method for PD recognition. Cui et al adopted the image moments characteristic parameter of PD to analyze the surface discharge development process [7]. The data size of these methods is very large and the speed of data processing is slow, which is not suitable for online monitoring
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call