UHF Partial Discharge Localization Algorithm Based on Compressed Sensing

Partial discharge (PD) localization is an effective way to detect the insulation problem of power equipment. Considering the environmental adaptability and low hardware cost, a novel ultrahigh frequency (UHF) PD localization algorithm based on received signal strength indication (RSSI) fingerprint is proposed. The methodology includes two stages: in the offline, the RSSI fingerprint map is established by filed test and the BP neural network is trained. Particle swarm optimization (PSO) algorithm is also adopted to choose the initial weight of the BP neural network. In the online stage, the PD sources localization which consists of two steps is deployed. Firstly, a preliminary localization achieved by trained BP neural network. Secondly, more accurate localization result is obtained using compressive sensing (CS) algorithm. Furthermore, since the UHF signal is sensitive to the layouts change of detection area, it is necessary to update the fingerprint map timely from practical application point of view, which is also a time-consuming process. To reduce the workload of fingerprint map updating, a reconstruction algorithm based on CS theory is proposed, by which the fingerprint map can be rebuilt by only a subset of original fingerprint data. A filed experiment in the substation is performed and the results show that, the average localization error of the CS algorithm is 0.89 meter and 90.4% localization errors are less than 2 meters. The results proved the accuracy and effectiveness of our proposed localization algorithm.