Fault Diagnosis of RV Gearboxes Based on Attention Mechanism
Traditional approaches to fault detection in RV reducers using deep learning heavily rely on manually defined feature extraction and various classifiers.However,these methods have limitations in capturing complex fault patterns and variations,and they exhibit poor generalization when dealing with different fault types that require re-extraction of features.To address this issue,this study propos-es the integration of attention mechanisms to adaptively weight and focus on key information in the in-put data,improving fault detection accuracy and generalization.Specifically,attention mechanisms are in-corporated into both convolutional neural network(CNN)and long short-term memory(LSTM)models to detect and identify vibrations in the helical gears of RV reducers based on collected signal data,aiming to enhance fault detection performance.Comparative tests on the accuracy of the models on a test dataset demonstrate that the attention mechanism significantly improves the accuracy of the net-work.This indicates that attention mechanisms can effectively extract features related to faults in the helical gears of RV reducers,enabling more accurate fault detection.
万方数据
维普
