In order to make better about ball bearing fault diagnosis, multiple sources and multiple methods about integration of diagnostic techniques are proposed based on ball bearing of vibration signal considering three kind of methods, including the wavelet analysis, the delayed correlation-envelope technique and Hilbert-Huang Transform (HHT), which uses three acceleration sensors test bearing seat acceleration to get multiple sources of vibration data. Using three methods to get feature values of the ball bearing fault and studying the nine kinds of integration programs, the different methods and data fusion diagnosis are discussed with support vector machine (SVM) for a feature fusion. Then through experimental verification and comparison of integration plans, the feasibility of the integration and effectiveness of diagnostic methods are shown.