Fault Diagnosis of High-Voltage Circuit Breakers via Hybrid Classifier by DS Evidence Fusion Algorithm.

Accurate and timely fault diagnosis is of significance for the stability of high-voltage circuit breaker (HVCB), which plays an important role in ensuring the safety of the power system. Current fault diagnosis techniques generally depend on an individual classifier. In this study, combining support vector machine (SVM) with extreme learning machine (ELM) by Dempster-Shafer (DS) evidence fusion algorithm, we propose DS_SE, a hybrid classifier for clearance joint fault diagnosis of HVCB. At first, through variational mode decomposition (VMD), the energy distribution is extracted as the feature value of vibration signals. Then DS evidence fusion algorithm is proposed for fusing analysis of evidence from different sub-classifiers and sensors. Extensive evaluation of DS_SE based on a real Zn12 HVCB indicates that it can achieve better performance by fusing conflicting evidence.