CNNMamba and CNNBIMamba: Simple and effective Fault Diagnosis Models based on Mamba

To verify the applicability of Mamba in the field of fault diagnosis, we propose a simple fault diagnosis model MetaNet, which uses LSTM, BISTM, GRU, BIGRU, Mamba, and BIMamba as its SignalMixer components, respectively. Experiments on the Xi’an Jiaotong University gearbox dataset have shown that MetaNet based on Mamba and BIMamba outperforms other models, with accuracy rates of 98.16% and 99.15% on the test set. The research results indicate that Mamba is a promising mechanical fault diagnosis model with potential applications in other fields.