Refined time-shift multiscale dispersion Lempel-Ziv complexity to diagnose rolling bearing faults

The key to damage detection is whether fault features can be extracted effectively from raw signals. Hence, we propose an approach based on the refined time-shift multiscale dispersion Lempel-Ziv complexity (RTSMDLZC) to effectively extract fault features. First, the time-shift multiscale sequence constructed from the raw time series can obtain more fault information more effectively. Then, the refined method addresses the lacking of sizeable numerical fluctuation on a large scale and enhances the algorithm's stability. Simulation signals and two experimental cases verify the effectiveness and applicability of the RTSMDLZC. The results indicate that compared with other classic methods, the RTSMDLZC can extract bearing fault features more accurately and has better identification accuracy.