农用PMSM匝间短路故障诊断模型超参数影响研究OACHSSCD

Permanent magnet synchronous motors(PMSMs)are key components of electric drive systems in agricultural equipment,and their operating status directly affects the safety and reliability of equipment operation.Inter-turn short-circuit fault is one of the most common and difficult-to-detect faults in PMSMs.If not identified in time and effectively addressed,the fault may further deteriorate,leading to motor burnout and even safety incidents.Therefore,early diagnosis of inter-turn short-circuit faults is of great significance for ensuring the safe and stable operation of PMSMs.With the rapid development of deep learning in the field of fault diagnosis,higher requirements have been placed on hyperparameter tuning efficiency and diagnostic accuracy.However,traditional manual hyperparameter tuning is highly subjective and inefficient,making it difficult to meet these requirements.Considering that hyperparameter selection has a significant impact on model performance,this paper develops a PMSM inter-turn short-circuit fault diagnosis model based on a dilated convolutional neural network(DCNN),and systematically investigates,using a controlled variable approach,the influence of structural and training hyperparameters on the model training process and validation accuracy in PMSM inter-turn short-circuit fault identification.The results show that different hyperparameters differ significantly in both the degree and dimension of their influence,and can therefore be classified into three categories:major,minor,and stable hyperparameters.Specifically,major hyperparameters affect both the model training process and validation accuracy,minor hyperparameters mainly affect only one of these two aspects,and stable hyperparameters should be determined before model training.The findings provide a basis for subsequent hyperparameter optimization and help reduce computational cost while improving model training efficiency and diagnostic performance.