A Combined Acoustic and Vibration Fault Diagnosis Method for High-Voltage Circuit Breakers Based on VMD and IPSO-SVM

Mechanical faults in high-voltage circuit breakers threaten power system safety, with traditional single-signal methods suffering from incomplete information or noise sensitivity; this paper proposes a joint acoustic-vibration diagnosis method using Variational Mode Decomposition (VMD) and Improved Particle Swarm Optimization-Support Vector Machine (IPSOSVM), where denoised signals are decomposed by VMD to extract Hilbert marginal spectrum energy entropy from Intrinsic Mode Functions (IMFs) as fault features input into an IPSO-optimized SVM for diagnosis. Experiments on a ZN65 vacuum circuit breaker simulated three faults (rotor jamming, base looseness, refusal to operate), and after wavelet packet soft threshold denoising, joint acoustic-vibration features achieved $100 \%$ diagnosis accuracy via IPSO-SVM, outperforming single vibration ($\mathbf{9 7 . 5 \%}$) and acoustic signals ($\mathbf{9 0 \%}$). VMD alleviates mode mixing, Hilbert energy entropy captures fault-specific energy distributions, and IPSO improves SVM parameter optimization through chaotic initialization and dynamic inertia to enhance generalization; this method offers a precise diagnostic approach with future work focusing on dataset expansion and cross-validation.