Use of Higher Order Spectra in Condition Monitoring: Simulation and Experiments

Abstract

Abstract In the field of machine condition monitoring one can observe that a link exists between machine vibrations and its health condition, that is, there is a change in the machine vibration signature when machine faults occur. Damages occurring in machine elements are often related to non-linear effects, which may lead to non-linearities in the machine vibration. This paper concerns the study of some systems by means of techniques based on Higher Order Spectra (HOS). These techniques are particularly useful in the situation where only a single measurement sensor is available. If a process is Gaussian then HOS provide no information that cannot be obtained from the second order statistics. On the contrary, HOS give information about a signal’s non-Gaussianity. Since a Gaussian input passing through a linear system leads to a Gaussian output, assuming the signal as an output of a system with a Gaussian input, then HOS make it possible to analyse the structure of the output signal and to provide information related to the non-linearity within the system. A simple model is presented with the aim of showing the effectiveness of the normalised version of polyspectra in detecting different kinds of system non-linearities. HOS are used to interpret the signal structure and the system’s physical characteristics. Moreover, two experimental cases are presented. The HOS are applied to detect the presence of a fatigue crack in a straight beam and to analyse the vibration signal measured on a test bench for rolling element bearings. Both third and fourth order spectra seem to provide a possibility of using HOS as a condition monitoring tool.