The changes of complexity in the performance degradation process of rolling element bearing

Rolling element bearing is not only the most important failure unit in rotary machinery, but it is also the most common, and its performance degradation assessment has been proposed to realize equipment’s near-zero downtime and maximum productivity. Therefore, exploring effective indices is crucial. Rolling element bearing’s vibration signals often contain complex, non-stationary and nonlinear characteristics. In this work, the changes of complexity, which are measured by two nonlinear methods: correlation dimension and approximate entropy, in the performance degradation process of rolling element bearing based on simulation and bearing accelerated life test data have been studied. Results show that the complexity decreases with the development of obvious defects, and that correlation dimension and approximate entropy have different sensitivities to different performance degradation stages, with the different ability of reflecting the first degradation stage being explained by their different computation processes.