A Probabilistic Model-Aided Failure Prediction Approach for Spring-Type Operating Mechanism of High-Voltage Circuit Breakers

To avoid failures in circuit breakers (CBs) and to extend the lifetime of these critical components, condition-based-maintenance has been increasingly requested by utilities to enable them to efficiently manage their assets. The origin of the most failures in CBs is the operating mechanism. Travel curve (TC) could effectively reveal the condition of the operating mechanism. However, the measurement of a TC profile is not simple in all CBs. This paper presents the impacts of common failure modes of CBs on TC profiles, and proposes a new model-aided approach to simulate the behavior of the operating mechanism with coupling the model-based and rule-based approaches. The simulation results along with experiments conducted on 72.5 kV SF6 CBs are organized into a fuzzy-probabilistic approach through maximum likelihood and interacting multiple models (IMM) to precisely predict the condition of CB and to detect intelligently the cause of the failure. For this purpose, the CB condition has been categorized into three modes based on its operating speed: normal (about 3 m/s), faulty-1 (less than the normal range), and faulty-2 (more than the normal range). The mode variations of the CB have been estimated via IMM in each operation. The proposed approach in prediction of the failures and cause(s) prior to their occurrence has been verified against experiments.