Bearing fault diagnosis based on HOA-FMD and SA-FNO model

In the past few years, deep learning has demonstrated significant potential in the area of signal processing and fault diagnosis, but how to effectively extract fault features and enhance the robustness and accuracy of the model is still a pressing issue to address. Targeting the issues of inadequate feature extraction and insufficient model generalization ability in bearing fault diagnosis, this paper presents a new fault diagnosis method based on Hippo optimization algorithm optimized Feature Mode Decomposition (HOA-FMD) and Fourier Neural Operator combined with Self-Attention mechanism (SA-FNO). Firstly, the original signal was decomposed by HOA-FMD to extract multiple mode components, and the five modes most related to the original signal were selected as feature inputs by combining the cosine similarity. On this basis, this paper uses a deep learning model based on SA-FNO to improve the capacity to identify key features. In the process of model training, the accuracy and robustness of fault diagnosis are further improved by multi-model integration and weight optimization strategy. Finally, to verify the validity of this method, it uses the CWRU data set and compares with three traditional algorithms. The results indicate that the presented method has achieved great performance improvement and higher accuracy.