Application of independent component analysis to fault diagnosis of gearbox

A signal preprocessing method based on independent component analysis(ICA) is put forword for fault diagnosis of a gearbox. The ICA algorithm based on mutual information minimization is derived and verified by mixed deterministic signals. Vibration signals measured from the gearbox are decomposed by ICA and it shows that the source signals have obvious fault characteristics. The fault information is enhanced when the ICA technology is applied to the fault diagnosis of the gearbox, which changes the traditional way of enhancing fault information mainly by reducing measurement noise, and provides a method for the diagnosis of weak faults.