Degradation Tracking of Rolling Bearings Based on Local Polynomial Phase Space Warping

The condition monitoring of rolling bearings has received much attention in prognostics and health management. Real-time monitoring of the bearings’ degradation provides vital information for planned maintenance of machinery. However, tracking this degradation is challenging due to the hidden nature of the damages. In this article, the local polynomial phase space warping (LPPSW) algorithm is proposed to monitor the damages of bearings with high accuracy. Damages change the parameters of bearing dynamical systems and warp the trajectory in reconstructed phase space (PS). In the LPPSW algorithm, the kernel function is applied to weigh the local nearest neighbor points in the reconstructed PS. Meanwhile, the quadratic polynomial model is designed to predict the reference PS trajectory. The trajectory error between the reference PS and the damaged PS is then computed by the LPPSW. Finally, the degradation is tracked in real time. Numerical simulations and run-to-failure experiments of bearings are employed to demonstrate the effectiveness of the LPPSW. The experimental results demonstrate that the LPPSW reveals a more obvious degradation trend when compared with PS warping method and commonly used damage indicators. The proposed LPPSW algorithm improves damage monitoring capabilities while boosting the predictive maintenance of bearings.