The vibration signal of rolling bearing has non-stationary and nonlinear characteristics. In order to apply the advantages of deep learning recognition of 2-D images to the fault diagnosis of rolling bearings, a multi-layer nested scatter plot-convolutional neural network (NSP-CNN) rolling bearing fault diagnosis model is proposed. The model uses fast Fourier transform to obtain the frequency spectrum of the vibration signal in different directions, and divides the frequency bands. After that, the signals of different bandwidths are given different colors to highlight the fault information of the rolling bearing. Finally, the combined NSP The optimized CNN model is input to the feature map to realize fault diagnosis. The results show that the model can achieve high diagnostic accuracy when diagnosing rolling bearing faults.