Identification of acoustic signals of corona discharges under different contamination levels using wavelet transform

Abstract

Partial discharges (PDs) emit energy in several ways, producing electromagnetic emissions in the form of radio waves, light and heat, and acoustic emissions in the audible and ultra-sonic ranges. These emissions enable us to detect, locate, measure, and analyse PD activity in order to identify faults before the development of failures, because once present, the damage caused by PDs always increases, leading to asset losses, outages, protection-system failure, disaster, and huge energy losses. Therefore, it is of great importance to identify different types of PDs and to assess their severity. This paper investigates the acoustic emissions associated with corona discharge (CD) from different types of sources in the time domain, and based on these it is used to detect, identify, and characterize the acoustic signals due to CD activity. Which usually takes place on polluted glass insulators used in high-voltage transmission lines and hence to differentiate abnormal operating conditions from normal ones. A laboratory experiment was conducted by preparing prototypes of the discharge. This study suggests a feature extraction and classification algorithm for CD classification. A wavelet signal processing toolbox is used to recover the CD acoustic signals by eliminating the noisy portion and to reduce the dimensions of the feature input vector. The proposed model is proven to characterize the PD activity with a high degree of integrity, which is attributed to the effect of the wavelet technique. The test results show that the proposed approach is efficient and reliable.