In recent years

fault diagnosis methods based on neural networks have shown great advantages in the accuracy and efficiency of diagnosis. However

the exponentially increasing number of model parameters limits the application of neural networks in engineering practice. Aiming at this problem

this paper proposes an improved MobileNet network based on one-dimensional convolutional neural network for fault diagnosis of rolling bearings. The improved network can be directly applied to one-dimensional vibration signals

effectively reducing the requirements of system hardware resources and realizing lightweight deployment of the network; The proposed method is validated using the Western Reserve University bearing dataset and the QPZZ-Ⅱ fault simulation test bench dataset. The accuracy of the model proposed in this paper is more than 99.8%

and the number of parameters is 1/2 of the standard convolutional neural network. The method proposed provides a new way for realizing intelligent diagnosis in light-resource embedded systems.