Optimization of KELM circuit breaker fault diagnosis based on EEMD and SOA
Hao Guo, X. Zhang · IET conference proceedings. · 2026
To address the challenge of determining critical parameters in the Kernel Extreme Learning Machine (KELM) for circuit breaker fault diagnosis, which significantly impacts diagnostic accuracy, this study proposes an optimized KELM approach for circuit breaker fault diagnosis based on Ensemble Empirical Mode Decomposition (EEMD) and the Snake Optimization Algorithm (SOA). First, to address the issues of strong nonlinearity and high noise content in circuit breaker vibration signals, EEMD is employed for noise reduction and feature extraction, decomposing the original signal into a series of intrinsic mode functions (IMFs) and establishing a feature vector matrix. To address the issue of high feature vector dimension, Kernel Principal Component Analysis (KPCA) is used for dimension reduction. This paper utilizes the KELM method for fault diagnosis. To address the issue of the significant impact of hyperparameters on the fault diagnosis accuracy of the KELM method, this paper proposes to optimize its parameters using SOA and establishes a SOA-KELM model to perform circuit breaker fault diagnosis. Experimental results show that the fault diagnosis accuracy of the proposed method is 98.75%, significantly higher than that of the Genetic Algorithm (GA) and Dung Beetle Optimizer (DOB), providing a new approach for circuit breaker fault diagnosis.