Research and Application of BSS Algorithm on The Gearbox Fault Diagnosis Based on The MMI Criterion

This paper presented a kind of blind source separation (BSS) technology and applied it into the gearbox fault diagnosis through the blind mixing signal separation. The algorithm based on the natural gradient fixed step-length was used to calculate the statistical independent source signal estimate value, and successfully extracted the fault information according to the separation signal power spectrum based on the minimum mutual information (MMI) criterion. The gearbox fault condition can be diagnosed effectively through the experiment proved, which provided a new method to the mechanical equipment fault diagnosis and running state monitor.