Adaptive construction of multiwavelet and research on composite fault diagnosis of rolling bearing

The diagnosis of composite fault occurred in mechanism is a difficult challenge at present. It is hard to diagnose the composite fault exactly and comprehensively due to the diversity and influence of faults. Multiwavelet has two or more base functions and many excellent properties. It might match two or more features of composite fault. The theory of multiwavelet supplies a reference for the solution of composite fault. The paper proposes an adaptive construction of multiwavelet using symmetric lifting. The kurtosis is selected to be the object of optimization. A new multiwavelet is constructed for the signal and used to decompose it redundantly. The distribution of relative energy for the characteristic frequency is computed. The composite fault is diagnosed by the sensitive frequency bands selected. Taking the composite faults of rolling bearing as examples, the effectiveness of the proposed method is proved.