High-voltage circuit breaker fault detection based on improved complete ensemble empirical mode decomposition with adaptive noise and firefly algorithm-optimized least squares support vector machine

Abstract

To address the challenges of difficulty in extracting fault features and low detection accuracy for high-voltage circuit breakers, this study proposes a novel method for fault detection based on the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) and the Firefly Algorithm (FA) optimized Least Squares Support Vector Machine (LSSVM) for high-voltage circuit breaker fault detection. First, the ICEEMDAN method is used to process the circuit breaker vibration signals, decomposing them into a series of intrinsic mode function (IMF) components. The energy entropy of the obtained IMF components is calculated as the fault feature vector of the circuit breaker. To address the issue of high-dimensional fault feature vectors, kernel principal component analysis (KPCA) is employed for dimensionality reduction and feature selection. The selected features are then used for fault detection via the LSSVM method. To address the issue of the significant impact of the core parameters on the classification performance of the LSSVM method, FA is employed for parameter optimization to determine the optimal parameter combination, and an FA-LSSVM fault detection model is established. Test set samples are input into the FA-LSSVM model for fault detection. Experimental results show that the fault detection accuracy of the proposed method for high-voltage circuit breakers reaches 99.17%. To validate the effectiveness of the proposed method, it is compared with the particle swarm optimization algorithm (PSO) and dung beetle optimization (DBO) algorithms. The proposed method achieves a fault detection accuracy rate that is 3.34% higher than the PSO-LSSVM method and 2.5% higher than the DBO-LSSVM method. The experimental results demonstrate that the proposed method has strong theoretical and engineering application value for high-voltage circuit breaker fault detection.