Support vector machine based fault diagnosis using trip and close coil current of circuit breakers

Circuit breakers (CB) play a key role in ensuring the reliability and safety of power systems. Therefore, an online condition monitoring system dedicated to circuit-breaker is critical. The key to such a monitoring system is to acquire valuable signals and use appropriate methods to extract useful features, after which relationships between features and faults should be built. Practically, a large proportion of CBs´ irregular operations are caused by faults occurring in the trip and close coil circuit (TCCC), whose duty is to receive instructions and to control operations of main CB contacts. The operation course of TCCC is well reflected by its trip and close coil currents, thus, the focus on the analysis of the latter could effectively assess the condition of TCCC. In this paper, the operation mechanism of TCCC is introduced and useful features embedding in current signals are extracted. Then the method of noise reduction using wavelet is introduced and quantities of fault simulation tests are performed followed by current signals acquirement and features extraction. Finally, support vector machine (SVM) is employed in the diagnosis stage, which is proved to be effective in CB fault classification.