Current-based trip coil analysis of circuit breakers for fault diagnosis

Trip coil current trace analysis is commonly used in the condition monitoring of circuit breakers. Its effective implementation enables an unobtrusive method of diagnosis, reducing costs and increasing safety. Conventional methods of analysis rely on the extraction of knowledge-driven, time-based features from the trace, followed by diagnosis through an expert system. Many utility companies rely on the automated feature extraction capabilities provided within the trip coil current recorder. This paper highlights markers of potentially poor circuit breaker health missed when relying solely on such time-based analysis. A supplementary, data-driven method focusing on identifying such cases using machine learning techniques is introduced and demonstrated in this paper. Additionally, attention is drawn to the susceptibility of incorrect feature extraction by the recorders when subject to some of the explained current-based abnormalities.