Blind Separation of Partial Discharge Mixing Signals
Chenguo Yao · Gao dianya jishu · 2010
The complexity of internal structure and the polymorphism of insulation defect types in gas insulated switchgear (GIS) lead to multiform insulation defects possibly existing in GIS under long-term operating voltage. We studied the separation of partial discharge (PD) mixing signals produced by two types of insulation defects and put forward the SONS separation algorithm which was suitable for the non-stationary source signals. The separation was implemented for simulating PD mixing signals by single PD source signals incited by single insulation defects and actual PD mixing signals in GIS model. Two evaluation parameters including performance index (PI) and signal interference ratio (SIR) were employed to evaluate its separation effect. The parameters via separation indicate the SONS algorithm can effectively separate mixing PD signals and regain single PD source signals. Furthermore,the SONS algorithm can be used to separate PD mixtures of more than two types of insulation defects,which is convenient to the subsequent defect identification of multiple insulation defects in GIS.