The application of time–frequency reconstruction and correlation matching for rolling bearing fault diagnosis

Abstract

There are still some remaining issues for time–frequency distribution application in rolling bearing fault diagnosis, such as noise suppression and resolution improvement. In this paper, we proposed a novel time–frequency correlation matching and reconstruction method to enhance the ability of rolling bearing fault identification. Firstly, we use the optimal simulated bearing fault signal to obtain the matching template through time–frequency distribution. Then, correlation matching operation is conducted between the obtained matching template and the original time–frequency distribution of analyzed signal. Finally, the original time–frequency distribution is reconstructed with the correlation coefficients and matching template using the template reconstruction algorithm. The reconstructed time–frequency distribution has inherited the capability of matching template in noise suppression, and can reveal the fault impulses of interest in a unified scale. The effectiveness of the proposed method has been proved by experimental result.