An empirical re-sampling method on intrinsic mode function to deal with speed variation in machine fault diagnostics

Order tracking has proved to be effective in dealing with the effects of speed variation in the analysis of rotating machinery vibration signals. To implement traditional order tracking in practice requires rotational speed information. However, it may be difficult in some cases to mount an appropriate monitoring device to obtain reliable speed information. In this paper, a novel empirical re-sampling of intrinsic mode functions obtained from empirical mode decomposition is explored, so that the approximation of order tracking effects without rotational speed is possible. At the same time, the newly introduced intrinsic cycle concept in the intrinsic mode function simplifies linking of the resultant spectra to signal variations, and is therefore beneficial for condition monitoring of rotating machines. In the paper the rationale behind the technique is first explained. Secondly, the effectiveness of the technique is demonstrated on a dynamic gear simulation model. Lastly, the technique is applied to experimental data from a gearbox test rig. Both the simulation and experimental studies corroborate the usefulness of the proposed technique.