Find in Library

In order to improve fault diagnosis accuracy of gear， an improved feature selection （IFS） method is proposed based on share feature selection（SFS） which is usually used. In the IFS， an independent fault feature set is established for binary fault type based on the characteristics of the two fault types of gears which used to instead unified fault feature set for all fault types of gears. And then， the independent fault feature sets are identified by establishing multiple support vector machines with multi binary classification support vector machine and the diagnosis results are obtained. The example of gear fault diagnosis shows that the improved feature selection method eliminates the interference of useless feature， the diagnosis accuracy is improved and has certain advantages.