Mechanical Fault Diagnosis of Circuit Breaker Based on Improved BREMD and ELM

Abstract

In order to accurately and quickly obtain the fault information from the circuit breaker vibration signal, a novel feature extraction method based on improved bandwidth restricted empirical mode decomposition (IBREMD) is proposed, and then the extreme learning machine (ELM) is employed for fault diagnosis. Since it is difficult to determine the optimal bandwidth frequency in traditional bandwidth restricted empirical mode decomposition, the introduced IBREMD approach adopts an optimization function to select the bandwidth restricted signal frequency, which is used during the process of empirical mode decomposition. In this way, the frequency resolving ability is dramatically improved, resulting in excellent feature extraction performance. The experimental results show that the diagnosis accuracy is 98.3%, which promise to be effective in terms of fault diagnosis of circuit breaker.