Enhanced Mechanical Fault Diagnosis of High-Voltage Circuit Breakers Using a Multi-Strategy Improved Dung Beetle Algorithm and Support Vector Machine

High-voltage circuit breakers (HVCBs) are critical switching devices whose mechanical reliability directly affects the safe and stable operation of power systems. However, accurate fault diagnosis of HVCBs remains challenging due to complex mechanical structures, nonlinear vibration characteristics, and sensitivity to parameter selection in data-driven models. To address these issues, this paper proposes an enhanced me-chanical fault diagnosis method based on a multi-strategy improved dung beetle op-timization–support vector machine (MIDBO-SVM) framework. First, mechanical vi-bration signals under four typical operating conditions of HVCBs are collected, and discriminative frequency-domain features are extracted using the fast Fourier trans-form. To overcome the limitations of conventional SVMs in parameter tuning, a mul-ti-strategy improved dung beetle optimization (MIDBO) algorithm is developed by integrating adaptive search mechanisms to enhance global exploration and conver-gence efficiency. The proposed MIDBO is then employed to optimize the penalty and kernel parameters of the SVM, yielding a robust and well-generalized fault diagnosis model. Experimental results demonstrate that the MIDBO-SVM model exhibits supe-rior convergence behavior and a stronger ability to escape local optima compared with standard optimization strategies. The proposed method achieves the highest diagnostic accuracy of 96.67% across multiple fault categories. Moreover, under imbalanced sample conditions, the MIDBO-SVM maintains high diagnostic accuracy and stability, effectively distinguishing different operating states of HVCBs. These results confirm the effectiveness and robustness of the proposed approach for mechanical fault diag-nosis of HVCBs.