Discharge Identification for SF6 Gas-Insulated System Applying Multi-Physical Information Fusion

The accurate identification and hazard assessment can effectively realize condition-based maintenance and equipment life cycle economic management. For this purpose, a novel discharge identification approach based on multi-physical information fusion is proposed considering practical applications in equipped engineering sites. Three typical discharge defect models are placed within the a SF6 gas-filled test chamber, where the high frequency current transformer (HFCT), ultra-high frequency (UHF) and acoustic emission (AE) methods-based sensors are placed for discharge pulse monitoring. Furthermore, the non-phase information of discharge pulse for each measurement are extracted by the developed host computer software, where the fingerprint characteristics, namely maximum pulse amplitude (MPA) and cumulative pulse amplitude (CPA) are further extracted. Finally, the extracted multi-physical characteristics are aggregated into feature matrix. The identification results indicated that diagnosis accuracy of the proposed model achieves an average accuracy of 99.33%, which demonstrates that the multi-physical information fusion can resulting in better diagnosis performance.