A Novel Method to Improve the Resolution of Envelope Spectrum for Bearing Fault Diagnosis Based on a Wireless Sensor Node

In this paper, an accurate envelope analysis algorithm is developed for a wireless sensor node. Since envelope signals employed in condition monitoring often have narrow frequency bandwidth, the proposed algorithm down-samples and cascades the analyzed envelope signals to construct a relatively long one. Thus, a relatively higher frequency resolution can be obtained by calculating the spectrum of the cascaded signal. In addition, a 50 % overlapping scheme is applied to avoid the distortions caused by Hilbert transform based envelope calculation. The proposed method is implemented on a wireless sensor node and tested successfully for detecting an outer race fault of a rolling bearing. The results show that the frequency resolution of the envelope spectrum is improved by 8 times while the data transmission remains at a low rate.