An Intelligent diagnosis method for rolling bearings based on Ghost module and adaptive weighting module

The vibration signals of rolling bearings are inevitably affected by noise and working conditions. The use of one-dimensional raw signals converted into images for rolling bearing fault diagnosis has achieved good results, but ignores the large model and diagnostic speed, and thus it is not suitable for practical fault diagnosis. To address this problem, we propose a method based on Ghost module and adaptive weighting module. The method utilizes Ghost modules and coordinated attention to make the model lightweight while improving the network's ability to extract features of the input data. Additionally, in order to effectively utilize the similar feature maps generated by convolution, an adaptive weighting module is proposed to further simplify the learning process and reduce the network training time. The validation using the datasets from Case Western Reserve University and the Association for Mechanical Failure Prevention Technology demonstrates the effectiveness of the proposed method. Under the same conditions, our method achieves 98.62% accuracy with only one-tenth of the parameters of classical neural networks. In noise environment simulations, our method exhibits strong noise immunity with 98.64% accuracy, along with robust diagnostic and generalization performance under various loads. Compared to the advanced fault diagnosis algorithms, our method boasts a 4% higher average accuracy and has its superiority in rolling bearing fault diagnosis.