Research On The Fault Diagnosis Method For Bearings And Gears Under Small Sample Conditions

As an important component of mechanical equipment,rotating machinery plays a vital role in many fields such as aerospace and industrial production,and how to ensure the safe and stable operation of these equipments has been a hot spot in the field of fault diagnosis.In recent years,deep learning has achieved some success in the field of rotating machinery fault diagnosis,however,most of the methods based on deep learning need to use a large amount of fault data to achieve accurate judgment of rotating machinery faults.In practical engineering applications,the amount of fault data collected is small due to many factors such as the operating environment and working conditions of rotating machinery.Therefore,the following research is done in this paper on how to construct a high-precision fault diagnosis algorithm in the case of small samples.(1)A fault diagnosis method based on triadic model was proposed.The method improves the training times of the model and alleviates the overfitting phenomenon of the model by constructing sample pairs in known fault samples,so that the model can effectively extract the features of faults,and then compare the similarity between the unknown fault sample features and the known fault sample features to achieve fault diagnosis.The validation results on bearing and gear datasets show that the method performs well in small-sample rotating machinery fault diagnosis.(2)A migration learning-based fault diagnosis method was proposed.The method introduces migration learning for practical engineering applications,where fault datasets often have fault data for multiple operating conditions and small data volume for a single operating condition,and introduces effective feature extraction from source domain samples and a small number of samples in the target domain,and defines multiple loss functions,so as to improve the fault recognition rate of the method.Experimental validation is performed with bearing and gear fault datasets,and the results show that the method has a good fault diagnosis recognition rate.(3)A fault diagnosis method based on TCNN-SMOTE was proposed.The method starts from the data side and performs data enhancement on small sample data to obtain sufficient data samples.The fault features of the samples in the small sample data set are first extracted using a feature extraction model,then the SMOTE algorithm is used to synthesize sample features over the extracted fault features,and finally the synthesized sample features are mixed with the extracted sample features,and next the fault classification model is trained using the mixed data set.The validation results on the bearing and gear fault datasets show that the fault diagnosis accuracy of this method after data augmentation is better than the traditional data augmentation method and the generative model GAN in the case of small samples.Figure 41 Table 18 Reference 73...