A method combining wavelet analysis and support vector machine which used in low-speed bearing fault diagnosis is presented.Firstly

every kind of bearing faults sample signal are collected from the experimental equipment and these signals are denoised by improved wavelet threshold.Then wavelet packet is used to decompose the denoised vibration signals

so the frequency band energy can be obtained and can be seen as eigenvector.Then the eigenvector can be taken as training sample of SVM multi-fault classifier for the intelligent trouble diagnosis.The result shows that there is a better efficiency by this combining method compared to traditional wavelet method.