Efficient method for location and detection of partial discharge in transformer oil by DOA estimation of circular array of ultrasonic sensors

Abstract

The electrical insulation failures in oil transformers are mainly occurs due to the inappropriate placing of Partial Discharge (PD) sources. In order to eliminate the insulation defects and also to locate the PD sources in an appropriate location, a new approach called circular array of ultrasonic sensors (CAUS) with various analysis is proposed. At first de-noise the PD signal from the CAUS using the fast independent component analysis (Fast ICA) algorithm. Secondly, the wide band signal from CAUS is converted into narrow band signal by using the total least square algorithm (TLS). Third, parse representation of array covariance vector (SRACV) technique is utilized to separate DOA (Direction of Arrival) in three directions from PD to CAUS. Finally, the PD sources are placed in an appropriate location by using the pitch and azimuth angles of those three DOAs and the exact coordination of three planes are calculated by using the particle swarm optimization algorithm. The simulation result demonstrates the effectiveness of proposed approach in terms of PD location in transformer oil.