采用Fisher判别法对GIS中局部放电类型的识别(英文)

To monitor the condition of online power apparatus accurately and provide appropriate guidance on their maintenance,a fundamental ultra-high frequency(UHF) database of partial discharges corresponding to different types of defects is presented for the observation of insulation state of gas insulated switchgear(GIS).In order to imitate the defects in a GIS online,five types of typical partial discharge(PD) sources(i.e.floating metal,protrusion,bouncing particles,void in spacer,and particle on the surface of solid insulation) were designed and fabricated.A wideband UHF sensor and an amplifier were applied to obtain UHF signals.Based on the characteristics of the five different defects,nine meaningful parameters which were independent of the phase of the applied voltage were extracted from the PD samples and then discussed.In this paper,Fisher discriminant analysis(FDA),a pattern recognition algorithm,was applied for the purpose of classifying the defects.In this way,there was no need to set complicated parameters and kernel function,and the total discrimination accuracy of test samples was 97.6%.It indicates that based on the nine especial parameters,the typical defects in GIS can be identified exactly by means of FDA.Hence the analysis method could be possibly regarded as a universal classification method.