Enhanced vibration monitoring using parametric modelling technique

A parametric modeling technique has been applied to the predictive maintenance of rotating machinery. The case, which is explained in detail, was taken from an experiment in which a one-step gearbox was run at about 150% of nominal load until failure occurred. During the experiment, vibration signals from the gearbox were measured. The novelty of this work is to analyze the residual signal obtained by computing the difference between the predicted and measured signal. A parametric modeling technique called autoregressive modeling is utilized in the prediction procedure. The analysis of the residual signal proved to be successful; the failure could be predicted easier and earlier than using traditional fast Fourier transform (FFT)-based methods