204 The Fusion Model of Dynamic Fault Diagnosis for Mechatronic Systems

There are changeful operating conditions, serious signal coupling of noise and fault in dynamic fault diagnosis of mechatronic systems. The fusion mode based on multi-scale wavelet transform, extended neighborhood rough set and classification of SVM is proposed. Firstly, Multi-scale wavelet transform is used to decompose vibration signal to every energy band according to frequency doubling. It can effectively extract fault feature. Secondly, extended neighborhood rough set based on generalized euclidean distance is used to eliminate noise and distribute weight of energy-band importance. Thirdly, the classification model of dynamic fault diagnosis is established based on SVC. Finally, the fusion method is applied to the fault diagnosis of gearbox and shows its accuracy and efficiency.