Fault Diagnosis of DC Circuit Breakers Utilizing Optimal VMD-Based Multi-Scale Features and SVM Classifier

Abstract

To address the complex electromechanical coupling characteristics of DC circuit breakers and the limitations of traditional single-scale features in characterizing incipient faults, this paper proposes a fault diagnosis method integrating optimized Variational Mode Decomposition (VMD) multi-scale features with a Support Vector Machine (SVM). First, the Mirage Search Optimization (MSO) algorithm is employed to synergistically optimize the critical parameters of both VMD and SVM, effectively overcoming the susceptibility of traditional optimization methods to local optima. Subsequently, the original vibration signals are decomposed into a series of Intrinsic Mode Functions (IMFs) via VMD, from which energy moments, kurtosis, and dispersion entropy are extracted to construct multi-scale feature vectors. Finally, these multi-scale composite features are fed into the optimized SVM for fault identification and classification. Experimental results demonstrate that the proposed method achieves a fault identification accuracy of 98.33%, significantly outperforming the control groups. The experimental data validate the advantages of employing multi-scale features and dual-parameter optimization, suggesting significant practical value for the complex fault diagnosis of DC circuit breakers.