The multi-source channel signals of the same monitoring point can provide more comprehensive feature information for accurate fault diagnosis,but the effective fusion of multi-source features is still challenging.To solve this problem,the coupled hidden Markov model (CHMM) was used to fuse the cyclostationary features of the dual-channel information effectively,that was,the features were extracted by fast spectral correlation (FSC),so as to provide effective multi-source fusion feature vector support for improving the accuracy of intelligent diagnosis of rolling bearings.The FSC analysis method was used to extract the features of the homologous dual-channel vibration signals of rolling bearings.The CHMM selected through parameter optimization was used to fuse the dual-channel homologous features to realize the intelligent diagnosis of rolling bearings.Through rolling bearing conventional fault experiments and full-life accelerated fatigue experiments,it is verified that the method can not only be used for intelligent classification of rolling bearing faults,but also for effective performance degradation evaluation of rolling bearings.Furthermore,the superiority of the described method is verified by comparative study.