383 / 2022-03-15 18:25:17

UHF Partial Discharge Localization Method Based on Time Fingerprint

partial discharge,localization method,Condition monitoring,time fingerprint

The existing partial discharge localization methods mainly include three types: time difference of arrival (TDOA), angle of arrival (AOA) and received signal strength indicator (RSSI). Among them, TDOA method uses the time difference of partial discharge signal reaching sensors to calculate the distance between partial discharge source and sensors, and then calculate the position coordinates of the source. However, the sampling accuracy of time difference is difficult to reach a high level, resulting in low localization accuracy; The principle of AOA method is similar to TDOA, but the coordinates of partial discharge source are calculated by the arrival angle of partial discharge signal, which is easy to be affected by environmental occlusion; The principle of RSSI method is to build the discharge strength fingerprint database of each position to match the discharge source coordinates, which environmental adaptability is strong and does not need high-frequency signal sampling, but the signal attenuation speed is not fixed, resulting in the instability of the accuracy. Therefore, a UHF partial discharge localization method based on time fingerprint is proposed in this paper, which combines the advantages of TDOA and RSSI. Firstly, the UHF sensor is used to collect the arrival time of partial discharge signal, and the signal arrival time difference between the sensors constitutes the time fingerprint of the measurement point. Then the time fingerprint database of the measured area would be got by the fingerprint of all measurement points. When partial discharge occurs, input the collected time fingerprint into the time fingerprint database and match it with neural network algorithm to obtain the localization result of partial discharge source. The experimental results show that in the measured area of 900 square meters, the average localization error of the proposed method is 1.78m, and about 60% of the localization error is less than 2m. The localization accuracy is high and the performance is stable. It can meet the application requirements of condition monitoring in substation.