OLTC Fault Diagnosis Method Based on Time Domain Analysis and Kernel Extreme Learning Machine

Abstract

<p>Aiming at the problems of limited feature information and low diagnosis accuracy of traditional on-load tap changers (OLTCs), an OLTC fault diagnosis method based on time-domain analysis and kernel extreme learning machine (KELM) is proposed in this paper. Firstly, the time-frequency analysis method is used to analyze the collected OLTC vibration signal, extract the feature information and form the feature matrix; Then, the PCA algorithm is used to select effective features to build the initial optimal feature matrix; Finally, a kernel extreme learning machine optimized by improved grasshopper optimization algorithm (IGOA), is used to handle the optimal feature matrix for classifying fault patterns. Evaluation of algorithm performance in comparison with other existing methods indicates that the proposed method can improve the diagnostic accuracy by at least 7%.</p> <p> </p>