Fault Diagnosis Model for Accessory Gearbox Based on Deep Transfer Learning

The accessory gearbox, as a critical component determining the performance of aircraft engines, directly influences the safety and reliability of the engine. However, in practical operation, the limited number of available fault samples due to changes in working conditions results in a small sample problem. This small sample issue can lead to underfitting of deep learning models. To address this challenge, this paper proposes a deep transfer learning approach based on the DenseNet-121 network for fault diagnosis. Experimental validation shows that with the application of transfer learning, the model achieves an accuracy of over 90% in identifying faults under different load conditions and speed conditions in small sample data. This method effectively enhances the accuracy and efficiency of fault diagnosis, providing a new perspective for addressing the small sample fault diagnosis problem in the accessory gearbox of aircraft engines under different conditions.