In view of the low accuracy and poor generalization ability of the conventional fault diagnosis technology,a convolutional neural network fault diagnosis method(STFT-CNN) based on short time Fourier transform (STFT) is proposed in this paper.In this fault diagnosis method,short-time Fourier transform(STFT) is applied to vibration signals under complex working conditions,and then the vibration data is trained and learned by convolutional neural network,and finally the fault diagnosis is carried out.In order to verify the effectiveness and feasibility of the proposed method in this paper,the proposed method is compared with convolutional neural network(CNN),support vector machine(SVM) and deep neural network among 15 types of rolling bearing faults including composite faults,In order to ensure the fairness of the experiment,the same rolling bearing data is used in the experimental comparison process.Experimental results show that the fault diagnosis accuracy of this method reaches 100%,which greatly improves the composite fault diagnosis accuracy of rolling bearing.