Abstract
Deconvolution methods utilizing time-domain sparsity as the objective function are widely adopted for machinery fault diagnosis due to their capability in adaptive filter design and attenuation of fault transmission. However, this approach presents several limitations, such as sensitivity to noises and reliance on prior knowledge. To address these challenges, this study introduces the Gini index (GI), derived from economics, as a replacement for time-domain sparsity in maximum envelope spectrum deconvolution. The GI is formulated as the objective function to mitigate the influence of impulse noises. Furthermore, the eigenvector algorithm is incorporated to optimize filter parameters in the proposed method. Finally, simulation and experimental results in a wind turbine generator demonstrate that the developed Gini-based maximum envelope spectrum deconvolution method has greater robustness against random impulse noises and non-fault harmonics.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call