Switchgear Partial Discharge Detection And Pattern Recognition Based On D-dot

Switchgear is widely used in electric power system. The safe operation of switchgear is of great significance for ensuring the safety of power grid operation and improving the quality of power supply. The fault in the switchgear can be quickly found by the monitoring of switchgear partial discharge, and thus speed up the efficiency of fault handling. As a kind of non-contact monitoring device, differential electric field (D-dot) sensor doesn’t affect the electrical performance of switchgear. Its volume and sensitivity can meet the requirements of partial discharge monitoring of high voltage switchgear. The mechanisms of switchgear partial discharge monitoring and testing method based on D-dot sensor are deeply studied in this paper.The PD cause and mechanism in switchgear are analyzed. Several typical PD types are summarized. The causing reason and discharge characteristics are described. On this basis, the principle of measurement of the D-dot probe is discussed, and then the related hardware is designed and manufactured.For the problem that traditional wavelet threshold algorithm can cause signal distortion, adaptive wavelet de-noising algorithm is studied. To solve the premature convergence problem of common threshold selection methods, a novel hybrid particle swarm optimization (HPSO) algorithm merging crossover mutation and chaos is proposed to obtain the global optimum thresholds. The de-noising results of simulative PD signals and the field PD signals are presented and the results show that hybrid particle swarm optimization adaptive wavelet threshold estimation (HPSOTE) has a fast convergence rate and an effective global optimization ability than the others, which significantly improved the credibility of the results and algorithm calculation speed. The HPSOTE can raise the SNR of signals and has good value in practical PD online monitoring.Switchgear partial discharge experiment platform is set up and the PD signal is measured by the probe and impulse current method at the same time. The probe’s reliability is verified by comparing the relationship of amplitude energy between PD signal and impulse current method. In addition, the measured local signal spectrum diagram at different detection ranges is studied. The relationship between high and low frequency component attenuation of signal and its propagation distance is analyzed.Last, the PRPD calculation and characteristic values identification is made. Based on the PD spectrogram, the discharge characteristic of the typical PD defects is analyzed. It can be concluded that the PRPD and characteristics are different at different discharge types, and thus can be used to make PD pattern recognition.