An Improved Variational Mode Decomposition with Adaptive Initial Center Frequency for Bearing Fault Feature Extraction

Variational mode decomposition (VMD) is extensively utilized in the field of industrial signal processing due to its superior anti-mixing and denoising capabilities. However, in practice, the performance of VMD is significantly influenced by the initial center frequency <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\omega^{1}$</tex>. An improper setting of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\omega^{1}$</tex> can significantly reduce the stability of decomposition, resulting in the loss or masking of feature information by noise. To address the above issue, an improved VMD algorithm is proposed in this work. Our algorithm redefines <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\omega^{1}$</tex> based on the narrowband characteristics of intrinsic mode function (IMF). Under the new definition, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\omega^{1}$</tex> can be adaptively adjusted based on the signal spectrum processed by Wiener filtering. This adjustment significantly enhances the flexibility and anti-mixing performance of the original VMD. To validate our proposed method, comparative experiments were conducted using two sets of fault data from Case Western Reserve University (CWRU). The results demonstrate that our method is more effective and superior in extracting fault features.