A Novel Hybrid Fault Diagnosis Method Based on EWT-SA-PNN

In order to improve the efficiency and accuracy of fault diagnosis, a novel fault diagnosis method based on signal processing and artificial intelligence method is proposed in this paper, which mainly combines empirical wavelet transform (EWT) with probabilistic neural network (PNN). A reference frequency method is proposed to make the input signals contain more complete feature information, which solved the problem of incomplete features in low frequency fault signals in practical industrial process. Firstly, the signal reconstruction is realized by EWT, thus the feature information in the signal is strengthened. Then, according to the 15 features commonly used in the signal, several feature parameters with high sensitivity are selected as features through sensitivity analysis (SA). Finally, the signal features and corresponding tags of different states are input into PNN to realize fault diagnosis. The experimental results show that this method can diagnose bearing fault quickly and effectively, and it can be applied to wind power fault diagnosis.