Study on application of independent component analysis in mechanical vibration signals
Ying Jian-zhong · Machinery Design and Manufacture · 2012
The independent component analysis(ICA)plays an important role in extracting characteristics of mechanical vibration signals.First independence criterion and optimization algorithms were established based on information theory as a framework in it.Then an index for measuring separation performance of independent component analysis was given,which optimization algorithm built can extract mixed signals very well.Finally,mixed signal from two Gauss signals and a Gauss signal were simulated,which results indicate that,the extracting effect of flexible ICA algorithm is better than that of the stochastic gradient algorithm,and the signal analysis method is characterized with its better convergences and less errors.