Aircraft security enhancement: Rotating machinery fault diagnosis and health assessment using manifold learning and dynamic time warping

To enhance the ability of AHMS, a research on aircraft rotating machinery fault diagnosis and health assessment techniques is conduct, and an approach combining manifold learning and dynamic time warping (DTW) is present in this paper. First, the original nonlinear and nonstationary vibration signals are processed by wavelet packet decomposition, and the wavelet energies are extracted to act as fault features, which are high-dimensional. Then, manifold learning method is employed for dimensionality reduction to find the intrinsic fault features. Finally, based on the accurate fault features, DTW is introduced to determine the fault state and assess the heath degree. The results of fault diagnosis and health assessment for a self-priming centrifugal pump in the aircraft fuel injection system demonstrate the effectiveness of the proposed approach.