Application of Natural Gradient Algorithm for Blind Source Seperation to Vibration Signal Processing
Guohong Li · Fire Control and Command Control · 2011
In this paper several algorithms of blind source separation are introduced and a simulation experiment on the mixture signals is conducted,which validates these algorithms.When using different steps,the separation performances of several algorithms,such as Infomax,EASI and natural gradient algorithm,are also compared in this paper.An improved natural gradient algorithm based on minimum mutual information are used to seperate the vibration signals of aero-engine and to get eigenfrequency of each vibration source,which is an available approach to seperating mechanical vibration signal sources.