Rolling bearing fault diagnosis using Adaptive Morlet wavelet and NGA optimized SVM

A rolling bearing fault diagnosis method is proposed based on adaptive Morlet wavelet and NGA optimized SVM. Firstly, three signal components nearby the appropriate scale as characteristic signals are extracted by adaptive Morlet wavelet, and their Shannon energy entropy are calculated respectively to form the sample set as input vector of SVM, in order to train the 1-v-r SVM. Then, a new nuclear function of SVM is constructed, and the kernel function parameters are optimized in the SVM training process by NGA in order to improve the classification performance of SVM. Finally, the experiment is carried out with the noisy rolling bearing mechanical fault data to prove its reliability and veracity.