In this paper, a bearing fault diagnosis method based on transfer learning is proposed to solve the problem that the traditional fault diagnosis method is not satisfactory under multi-working conditions. First, the Transfer Component Analysis method is employed to transform the source domain and the target domain into the same space. Then annotation probability matrix is proposed for fault diagnosis. Finally, the proposed method is verified on the bearing data set of CWRU university, and the recognition accuracy is obviously higher than the traditional methods. It is worth noting that the proposed method does not need parameters tuning and is very simple.