Considering the facts that different measurement point information provide different sensitivity degree and vibration transmission make fault relate to each measuring point,the fault diagnosis model of vibration relevant information fusion were put forward. The method is to obtain the eigenvector of wavelet packet signal energy about the normal and 5 different fault states under 3 measuring points. The correlation coefficient of the energy characteristics were calculated for the same measuring point in different states and for different measuring points in the same state. By analyzing the characteristics of these two types of fusion information,the fault diagnosis model was established. In addition,the fault diagnosis model was inspected with 14 groups of signals to be detected,and the results were exactly consistent with the actual fault states. The support vector machine method was adopted to realize classification of 5 different fault states. The outcome was consistent with the actual fault states and it verifies the correctness of the established model of fault diagnosis. Analysis results show that the fault diagnosis model was established through the vibration relevant information fusion,and the model can reflect the characteristics of different fault states.By the model,the weak and the coupling fault were effectively diagnosed for complex gear transmission system.