Qiang Miao1,4, Lin Cong1 and Michael Pecht2,3
1 School of Mechanical, Electronic and Industrial Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan 611731, People's Republic of China
2 Center for Advanced Life Cycle Engineering (CALCE), University of Maryland, College Park,
MD 20742, USA
3 Center for Prognostics and System Health Management, City University of Hong Kong, Hong Kong
Complex systems can significantly benefit from condition monitoring and diagnosis to optimize operational availability and safety. However, for most complex systems, multi-fault diagnosis is a challenging issue, as fault-related components are often too close in the frequency domain to be easily identified. In this paper, the interpolated discrete Fourier transform (IpDFT) with maximum side lobe decay windows is investigated for machinery fault feature identification. A novel identification method called the zoom IpDFT is proposed, which combines the idea of local frequency band zooming-in with the IpDFT and demonstrates high accuracy and frequency resolution in signal parameter estimation when different characteristic frequencies are very close. Simulation and a case study on rolling element bearing vibration data indicate that the proposed zoom IpDFT based on multiple modulations has better capability to identify characteristic components than do traditional methods, including fast Fourier transform (FFT) and zoom FFT.
Keywords: prognostics and health management, interpolated DFT, zoom IpDFT, Fourier transform, characteristic component identification
Full text is available from the publisher to CALCE consortium members.