The fault diagnosis of the equipment can ensure the smooth operation of the equipment. Especially, the structural abnormality of the equipment will affect the stability and reliability of the equipment. In this paper, a new method of signal fusion is proposed, which fuses the symptom parameters of multi-source vibration signals. The optimal data fusion model can effectively improve the reliability of data fusion algorithm, and sensitive symptom parameters are selected by the identification index (DI). In addition, a hybrid principal component analysis method (HPCA) is proposed. The least squares projection method is used to pre-process sensitive symptom parameters, which can effectively enhance the correlation of the symptom parameters. To recognize the fault type exactly and realize visualization, the principal component analysis algorithm is applied to reduce the dimension of the pre-processed symptom parameters, and precise diagnosis is achieved by analyzing the Mahalanobis distance of the integrated parameter. The experimental results show that the combination of signal fusion and hybrid principal component analysis can accurately identify the fault types.