Noise removal in Vold-Kalman order tracking based on independent component analysis

Vold-Kalman order tracking (VKOT) is a well known OT scheme with the advantages for beat-free order extraction of close or crossing orders in non-coherently running multiple shaft systems. However, the decoupling ability is often limited by both heavy computation and deficiency in information about rotary speeds or frequencies of complex mechanical systems. A new approach for removing crossing noise in VKOT is proposed in this paper to overcome these problems in traditional approach for crossing-noise decoupling. In the presented approach, independent component analysis (ICA) has been employed to remove the noise by the separation of independent components (ICs) before performing VKOT. The ambiguities of ICA are also solved in our scheme, which make it feasible to condition monitoring and fault diagnosis applications of rotating machinery. The approach is applied to simulation experiments. Analysis results give positive support to the presented approach.