Skip to Main Content
Partial-discharge (PD) detection is an important method to evaluate the insulation condition of metal-clad apparatus. Nonintrusive sensors, which are easy to install and have no interruptions to operation, are preferred in onsite PD detection. However, its accuracy often degrades due to the interferences in PD signals. In this paper, a PD extraction method that uses entropy-based time-frequency (TF) analysis is introduced. Entropy, which is a measure of disorder, is an ideal tool to extract PD or pulse-like interferences since the disorder of local TF spectrum increases when large-amplitude TF coefficients are contained. According to the characteristics of nonintrusive sensors and the distribution of entropy, the TF spectrums of PD and interferences are first separated. Then, the PD signal and interferences are recovered via inverse TF transform. The denoised results of some PD data as well as interferences demonstrate that the combination of entropy with TF analysis can discriminate PDs from interferences with a different TF entropy spectrum.